02 - Explore, Select & Filter - Python

Explore, Select & Filter - Python

Quote

“If we have data, let’s look at data. If all we have are opinions, let’s go with mine.”

— Jim Barksdale

Summary

Uses real EuroStoxx 50 OHLCV, dimension, and scoring tables to show how Pandas and Polars preview data, inspect structure and completeness, select rows and columns, and build reproducible filter logic for exploratory analysis and profiling.

Head / Tail / Sample / Glimpse

• Preview row shape and representative values with head(), tail(), sample(), and Polars glimpse(), including seeded sampling for reproducible notebook output
• Establish the three working datasets up front: eurostoxx50_ohlcv, index_dim, and scores_daily

Shape / Describe / Info

• Inspect row/column counts, summary statistics, dtypes, non-null counts, memory usage, schema, and transposed column previews with Pandas .info() / .describe() and Polars df.schema, null_count(), and estimated_size()
• Contrast Pandas’ numeric-default describe() with broader Polars summaries and Pandas workarounds such as include="all"

Value Counts / Unique / N-Unique / Null Inspection

• Profile categorical frequency, cardinality, duplicate keys, and missingness with value_counts(), unique(), n_unique(), .isna() / .is_null(), and null-count summaries
• Show how null behavior differs across libraries, including Pandas dropna=False requirements and NaN-vs-null pitfalls

Scores / Dimension / Profiling Strategies

• Profile the scores_daily table for sparse metrics, ranked score columns, and null-heavy analytical fields before downstream filtering
• Validate the index_dim dimension table for join readiness, categorical consistency, and key uniqueness, then generalize the profiling workflow into a repeatable checklist

Selecting Rows & Columns

• Select by label, position, name, dtype, and pattern with .loc, .iloc, bracket indexing, select(), select_dtypes(), and Polars selectors via polars.selectors as cs
• Contrast Pandas’ index-aware accessors with Polars’ index-free model, including slice semantics, row/column subsets, and how to emulate label filters in Polars

Filtering Rows

• Filter with boolean masks, df.filter(pl.col(...)), .query(), isin() / is_in(), between() / is_between(), string/date predicates, null checks, seeded sampling, deduplication, null dropping, sorting, and semi-join-style membership filters
• Compare conditional replacement patterns: Pandas where() / mask() versus Polars when().then().otherwise()

Operations and safety

• Warnings: Pandas filter() selects columns not rows, boolean logic requires & / | / ~ plus parentheses, NaN equality fails, .loc / .iloc slice semantics differ, deduplication depends on row order, and unseeded sample() is non-reproducible
• Recommendations: 7 practices covering profile-first workflow, early filter/select pushdown, cs selectors, validate= before join-dependent filters, is_in() over chained ORs, dropna=False in Pandas value_counts(), and sorting before head() / tail()
• Troubleshooting: 10 failure modes covering missing columns, unexpectedly empty filters, boolean-precedence TypeErrors, Series-vs-DataFrame selection mistakes, unordered uniques, and dtype mismatches in isin()

Glossary

head() / tail()

• Methods that return the first or last n rows of a DataFrame, usually used as the fastest preview of raw tabular values.

• They are the first inspection step in this note because they reveal row layout, obvious type issues, and whether the loaded data even resembles the expected dataset.

Preview is not profiling

A few visible rows can confirm column names and rough shape, but they do not reveal null rates, duplicate keys, or hidden outliers. Use them as the opening check, not the full diagnosis.

---

sample()

• A method that returns a random subset of rows from a DataFrame instead of the first or last rows in storage order.

• It matters here because random sampling helps catch localized anomalies that head() and tail() can miss, especially in time-ordered datasets.

Seed controls reproducibility

Without random_state= in Pandas or seed= in Polars, the sampled rows change every run. That breaks notebook reproducibility and makes debugging harder.

---

shape

• The (rows, columns) dimensions of a DataFrame.

• It is the fastest structural sanity check in the note because many downstream issues start with the wrong row count, unexpected width, or accidental empty results.

Cheap but high-signal

A shape mismatch often reveals upstream filter, join, or load errors before you inspect any actual cell values.

---

describe()

• A summary method that computes descriptive statistics such as count, mean, standard deviation, min, max, and percentiles for selected columns.

• It is used here as the first numeric quality screen for distributions, missingness, impossible ranges, and suspiciously constant columns.

Pandas hides non-numeric columns

Pandas excludes many non-numeric fields by default. If string, categorical, or datetime columns matter, use include="all" or inspect them separately.

---

info()

• A Pandas-only structural summary that reports column names, non-null counts, dtypes, and memory usage.

• It matters because it compresses schema validation and completeness checks into one glance before deeper selection or filtering work.

No Polars equivalent

Polars splits the same insight across schema, null_count(), and estimated_size(). You need multiple calls rather than one .info() method.

---

glimpse()

• A Polars display method that summarizes columns vertically, showing each column’s name, dtype, and preview values in a transposed layout.

• It is useful in this note for wide DataFrames where a normal horizontal preview hides important columns off-screen.

Display helper, not transformation

glimpse() is for inspection only. It does not change the underlying DataFrame or replace explicit schema and null checks.

---

value_counts()

• A frequency-count operation that reports how often each unique value occurs in a column.

• It is central to the note’s profiling workflow because it surfaces dominant categories, unexpected labels, and join-key quality issues quickly.

Pandas drops NaN by default

In Pandas, null-like values disappear unless you pass dropna=False. If you forget that flag, your cardinality analysis underreports missing data.

---

unique() / n_unique()

• Distinct-value operations: unique() returns the distinct values themselves, while n_unique() returns only the count.

• They matter for deciding whether a column is categorical, suitable as a key, or contaminated by unexpected duplicates.

Uniqueness is often unordered

Distinct values are commonly returned in arbitrary order. If the order matters for display or testing, sort the result explicitly.

---

Null / missing value

• The absence of a value in a column, represented as Arrow nulls in Polars and as NaN or nullable sentinels such as pd.NA in Pandas.

• Missingness is a core concern in this note because it changes counts, filters, deduplication, joins, and statistical summaries.

Equality does not find nulls

Missing values are not reliably detected with ==. Use .isna(), .notna(), .is_null(), or .is_not_null() instead.

---

Boolean mask

• A per-row sequence of True and False values used to keep rows whose condition evaluates to True.

• It is the basic row-filtering mechanism across the note, whether expressed directly in Pandas or through Polars expressions.

Use bitwise operators

Combine conditions with &, |, and ~, not Python and, or, or not. Each condition also needs parentheses to avoid precedence bugs.

---

.loc / .iloc

• Pandas accessors for label-based selection (.loc) and integer-position-based selection (.iloc).

• They matter because the note compares Pandas’ index-aware selection model with Polars’ index-free row/column access patterns.

Slice endpoints differ

.loc label slices include the right endpoint, while .iloc positional slices exclude it. Mixing the two models causes off-by-one selection errors.

---

Filter expression

• A boolean expression passed to Polars filter() and built from column expressions such as pl.col("close") > 50.

• It matters because Polars filtering is expression-driven and composes cleanly with selection, lazy execution, and optimizer pushdown.

Pandas filter() means something else

In Polars, filter() selects rows. In Pandas, filter() is mostly for column-label selection. The same method name points at different operations.

---

query()

• A Pandas method that filters rows using a string expression instead of explicit bracketed boolean masks.

• It appears in the note as an alternative filtering syntax for readable notebook code when conditions are simple and column names are expression-friendly.

String syntax has limits

query() is convenient, but it is still parsed text. Complex expressions, odd column names, or heavy refactoring can make explicit masks safer and clearer.

---

Selector / cs

• The Polars selectors API, imported as polars.selectors as cs, for choosing columns by dtype family, pattern, or set logic.

• It matters because the note uses selectors to avoid brittle hard-coded column lists when exploring changing schemas.

Separate module import

Selectors are not available automatically from the main pl namespace in the same way users often expect. Import polars.selectors as cs explicitly.

---

isin() / is_in()

• Membership tests that return a boolean mask indicating whether each value belongs to a provided list or set of candidates.

• They are used in the note for watchlist-style filters, semi-join-style membership checks, and compact alternatives to chained equality conditions.

Method names differ

Pandas uses .isin(), while Polars uses .is_in(). The underscore is easy to miss and is a common source of copy-paste errors.

---

between() / is_between()

• Range predicates that test whether values fall between two bounds.

• They matter because price ranges, date windows, and score thresholds are common filter shapes throughout exploratory work.

Boundary rules vary

Pandas is inclusive by default, while Polars exposes boundary control through closed=. If edge values matter, make the inclusion rule explicit.

---

where() / mask() / when().then().otherwise()

• Conditional replacement patterns that keep or replace values based on a boolean condition, with Pandas and Polars exposing different APIs.

• They matter because the note uses them to express row-dependent value logic without manually splitting and recombining DataFrames.

Pandas where and mask invert each other

where() keeps values where the condition is True, while mask() replaces values where the condition is True. Polars’ when().then().otherwise() is more explicit about branch direction.

---

drop_duplicates() / unique()

• Row-deduplication operations that keep one representative row from repeated values or repeated key combinations.

• They matter because profiling and filtering often depend on confirming whether keys are truly unique before joins or aggregations.

Kept row depends on order

If you keep the “first” duplicate without sorting deterministically, the survivor is only as stable as the incoming row order.

---

dropna() / drop_nulls()

• Row-removal operations that discard rows containing missing values, optionally limited to a subset of columns.

• They matter because many filters and comparisons behave differently once null-bearing rows are removed or isolated.

Row loss can be silent

Dropping nulls can remove far more data than intended, especially in wide tables. Always specify the critical subset when only certain columns matter.

---

Semi-join filter

• A filter that keeps rows from one table only when their key exists in another table, without importing the other table’s columns.

• It matters because the note contrasts Pandas membership-style workarounds with Polars’ native how="semi" join for table-driven filtering.

Useful for key membership

Semi-joins are clearer than full joins when you only care whether a key exists, not about bringing reference columns into the result.

---

Sort / sort_values()

• Row-ordering operations that arrange a DataFrame by one or more columns in ascending or descending order.

• They matter because many previews and top-N selections in the note only make sense after the data is explicitly ordered.

Parameter names differ

Pandas uses ascending=, while Polars uses descending=. The intent is the same, but the parameter names invert the phrasing.

---

import pandas as pd
import polars as pl
import polars.selectors as cs
import numpy as np
from pathlib import Path
import io
 
from IPython.display import display, Markdown
html_formatter = get_ipython().display_formatter.formatters['text/html'] # type: ignore
html_formatter.for_type(pd.DataFrame, lambda df: df.to_html())
html_formatter.for_type(pd.Series, lambda s: s.to_frame().to_html())
 
pl.Config.set_tbl_rows(100)
pd.set_option("display.max_rows", 100)
 
DATA = Path("../data")
 
# Core datasets
ohlcv_pd = pd.read_parquet(DATA / "eurostoxx50_ohlcv.parquet")  # 66K rows, daily OHLCV
ohlcv_pl = pl.read_parquet(DATA / "eurostoxx50_ohlcv.parquet")
dim_pd = pd.read_parquet(DATA / "index_dim.parquet")            # 169 rows, stock metadata
dim_pl = pl.read_parquet(DATA / "index_dim.parquet")
scores_pd = pd.read_parquet(DATA / "scores_daily.parquet")      # 466 rows, composite scores
scores_pl = pl.read_parquet(DATA / "scores_daily.parquet")
 
print(f"OHLCV: {ohlcv_pd.shape}, Dim: {dim_pd.shape}, Scores: {scores_pd.shape}")

OHLCV: (66355, 12), Dim: (169, 26), Scores: (466, 36)

Three datasets are used throughout: eurostoxx50_ohlcv (66K rows, daily OHLCV prices), index_dim (169 rows, stock metadata), and scores_daily (466 rows, composite scores with real nulls).

---

Head / Tail / Sample / Glimpse

The first step in data exploration is previewing rows. Both libraries provide .head() and .tail(). Polars additionally offers .glimpse() for a transposed column-by-column preview of data types and sample values.

Pandas | head() and tail()

display(Markdown("**First 5 rows (head):**"))
display(ohlcv_pd.head())

First 5 rows (head)

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

display(Markdown("**Last 5 rows (tail):**"))
display(ohlcv_pd.tail())

Last 5 rows (tail)

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
66350	64828	WKL.AS	2026-03-06	69.02	69.36	67.82	68.52	68.52	1143729	0.0	0.0	False
66351	66875	WKL.AS	2026-03-09	68.78	69.16	67.64	68.64	68.64	841503	0.0	0.0	False
66352	66876	WKL.AS	2026-03-10	68.80	69.16	66.34	67.16	67.16	1355645	0.0	0.0	False
66353	66877	WKL.AS	2026-03-11	67.50	69.60	67.02	67.22	67.22	1142531	0.0	0.0	False
66354	66929	WKL.AS	2026-03-12	67.00	67.54	66.28	67.32	67.32	210379	0.0	0.0	False

Pandas | sample()

display(Markdown("**Random sample of 5 rows:**"))
display(ohlcv_pd.sample(5, random_state=42))

Random sample of 5 rows

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
43053	38920	MUV2.DE	2023-03-29	320.0000	322.600	318.400	322.4000	290.3200	195809	0.0	0.0	False
56209	5772	SAP.DE	2022-11-03	95.9200	96.340	95.080	95.5100	91.9052	1424973	0.0	0.0	False
53515	11015	SAN.MC	2022-09-15	2.5975	2.686	2.597	2.6765	2.3373	70158349	0.0	0.0	False
6498	28954	AI.PA	2025-08-12	173.3200	174.440	172.800	173.6800	173.6800	415652	0.0	0.0	False
63527	24956	UCG.MI	2025-07-07	56.4600	57.350	56.440	57.3500	56.0468	4705567	0.0	0.0	False

Polars | head() and tail()

display(Markdown("**First 5 rows (head):**"))
display(ohlcv_pl.head())

First 5 rows (head)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

display(Markdown("**Last 5 rows (tail):**"))
display(ohlcv_pl.tail())

Last 5 rows (tail)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
64828	WKL.AS	2026-03-06	69.02	69.36	67.82	68.52	68.52	1143729	0.0	0.0	false
66875	WKL.AS	2026-03-09	68.78	69.16	67.64	68.64	68.64	841503	0.0	0.0	false
66876	WKL.AS	2026-03-10	68.8	69.16	66.34	67.16	67.16	1355645	0.0	0.0	false
66877	WKL.AS	2026-03-11	67.5	69.6	67.02	67.22	67.22	1142531	0.0	0.0	false
66929	WKL.AS	2026-03-12	67.0	67.54	66.28	67.32	67.32	210379	0.0	0.0	false

Polars | sample() and glimpse()

display(Markdown("**Random sample of 5 rows:**"))
display(ohlcv_pl.sample(5, seed=42))

Random sample of 5 rows

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
11529	SAN.MC	2024-09-18	4.511	4.5455	4.5065	4.5085	4.2785	16487238	0.0	0.0	false
35604	CS.PA	2025-10-14	39.34	40.27	39.25	40.18	40.18	3511125	0.0	0.0	false
63666	WKL.AS	2022-01-05	102.2	102.65	101.25	101.8	95.254	230509	0.0	0.0	false
33136	PRX.AS	2021-05-03	41.3654	41.737	41.0718	41.3746	40.8581	2177525	0.0	0.0	false
19417	SAF.PA	2024-07-12	204.2	204.8	201.3	204.8	202.5157	496739	0.0	0.0	false

display(Markdown("**Glimpse (transposed summary):**"))
# Transpose the glimpse into a horizontal table
cols = []
for line in ohlcv_pl.glimpse(return_type="string").strip().split("\n"):
    parts = line.split()
    if len(parts) >= 3:
        name = parts[0]
        dtype = parts[1]
        preview = " ".join(parts[2:])
        cols.append({"Column": name, "Type": dtype, "Preview": preview})
display(ohlcv_pl.head())

Glimpse (transposed summary)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

---

Shape / Describe / Info

Shape, summary statistics, column types, and memory usage are the core metadata inspection operations. Pandas provides .describe(), .info(), and .dtypes. Polars provides .describe(), .schema, and .dtypes.

Pandas | shape

for name, df in [("ohlcv", ohlcv_pd), ("dim", dim_pd), ("scores", scores_pd)]:
    print(f"{name:>10s}: {df.shape[0]:>8,} rows x {df.shape[1]:>3} cols")

ohlcv: 66,355 rows x 12 cols dim: 169 rows x 26 cols scores: 466 rows x 36 cols

Pandas | describe()

• Describe: Summary statistics: count, mean, std, min, max, quartiles.

display(Markdown("**Numeric summary:**"))
display(ohlcv_pd.describe())

Numeric summary

	id	open	high	low	close	adj_close	volume	dividends	stock_splits
count	66355.000000	66355.000000	66355.000000	66355.000000	66355.000000	66355.000000	6.635500e+04	66355.000000	66355.000000
mean	33179.733102	197.040520	199.364124	194.585782	197.034900	190.494909	5.942124e+06	0.011757	0.000172
std	19158.201385	363.150484	367.873829	358.011643	363.052047	359.635301	1.615619e+07	0.283142	0.022716
min	1.000000	1.601000	1.662800	1.584200	1.606600	1.201300	0.000000e+00	0.000000	0.000000
25%	16589.500000	29.789950	30.090000	29.470000	29.787450	28.143400	5.099855e+05	0.000000	0.000000
50%	33178.000000	70.700000	71.400000	69.890000	70.680000	63.141000	1.415896e+06	0.000000	0.000000
75%	49766.500000	185.990000	188.000000	184.000000	186.100000	175.253900	4.089299e+06	0.000000	0.000000
max	66930.000000	2926.000000	2957.000000	2813.000000	2839.000000	2802.938200	3.763915e+08	22.500000	5.000000

display(Markdown("**Include all dtypes:**"))
display(ohlcv_pd.describe(include="all"))

Include all dtypes

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
count	66355.000000	66355	66355	66355.000000	66355.000000	66355.000000	66355.000000	66355.000000	6.635500e+04	66355.000000	66355.000000	66355
unique	NaN	50	1331	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	2
top	NaN	ABI.BR	2026-03-12	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	False
freq	NaN	1331	50	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	66349
mean	33179.733102	NaN	NaN	197.040520	199.364124	194.585782	197.034900	190.494909	5.942124e+06	0.011757	0.000172	NaN
std	19158.201385	NaN	NaN	363.150484	367.873829	358.011643	363.052047	359.635301	1.615619e+07	0.283142	0.022716	NaN
min	1.000000	NaN	NaN	1.601000	1.662800	1.584200	1.606600	1.201300	0.000000e+00	0.000000	0.000000	NaN
25%	16589.500000	NaN	NaN	29.789950	30.090000	29.470000	29.787450	28.143400	5.099855e+05	0.000000	0.000000	NaN
50%	33178.000000	NaN	NaN	70.700000	71.400000	69.890000	70.680000	63.141000	1.415896e+06	0.000000	0.000000	NaN
75%	49766.500000	NaN	NaN	185.990000	188.000000	184.000000	186.100000	175.253900	4.089299e+06	0.000000	0.000000	NaN
max	66930.000000	NaN	NaN	2926.000000	2957.000000	2813.000000	2839.000000	2802.938200	3.763915e+08	22.500000	5.000000	NaN

Pandas | info()

# display hlcv_pd.info() as dataframe
 
info_df = pd.DataFrame({
    "Column": ohlcv_pd.columns,
    "Non-Null": [ohlcv_pd[c].notna().sum() for c in ohlcv_pd.columns],
    "Dtype": [ohlcv_pd[c].dtype for c in ohlcv_pd.columns],
})
display(info_df)

	Column	Non-Null	Dtype
0	id	66355	int64
1	symbol	66355	object
2	date	66355	object
3	open	66355	float64
4	high	66355	float64
5	low	66355	float64
6	close	66355	float64
7	adj_close	66355	float64
8	volume	66355	int64
9	dividends	66355	float64
10	stock_splits	66355	float64
11	is_filled	66355	bool

Pandas | dtypes

display(ohlcv_pd.dtypes)

	0
id	int64
symbol	object
date	object
open	float64
high	float64
low	float64
close	float64
adj_close	float64
volume	int64
dividends	float64
stock_splits	float64
is_filled	bool

Polars | shape

for name, df in [("ohlcv", ohlcv_pl), ("dim", dim_pl), ("scores", scores_pl)]:
    print(f"{name:>10s}: {df.shape[0]:>8,} rows x {df.shape[1]:>3} cols")

ohlcv: 66,355 rows x 12 cols dim: 169 rows x 26 cols scores: 466 rows x 36 cols

Polars | describe()

• Describe: Summary statistics: count, mean, std, min, max, quartiles.

display(Markdown("**Polars describe (all columns):**"))
display(ohlcv_pl.describe())

Polars describe (all columns)

statistic	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
str	f64	str	str	f64	f64	f64	f64	f64	f64	f64	f64	f64
count	66355.0	66355	66355	66355.0	66355.0	66355.0	66355.0	66355.0	66355.0	66355.0	66355.0	66355.0
null_count	0.0	0	0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
mean	33179.733102	null	2023-08-05 00:56:42.354005	197.04052	199.364124	194.585782	197.0349	190.494909	5.9421e6	0.011757	0.000172	0.00009
std	19158.201385	null	null	363.150484	367.873829	358.011643	363.052047	359.635301	1.6156e7	0.283142	0.022716	null
min	1.0	ABI.BR	2021-01-04	1.601	1.6628	1.5842	1.6066	1.2013	0.0	0.0	0.0	0.0
25%	16590.0	null	2022-04-20	29.79	30.09	29.47	29.7899	28.1461	509991.0	0.0	0.0	null
50%	33178.0	null	2023-08-03	70.7	71.4	69.89	70.68	63.141	1.415896e6	0.0	0.0	null
75%	49767.0	null	2024-11-19	186.0	188.0	184.0	186.1	175.2609	4.089463e6	0.0	0.0	null
max	66930.0	WKL.AS	2026-03-12	2926.0	2957.0	2813.0	2839.0	2802.9382	3.76391539e8	22.5	5.0	1.0

Polars | schema and dtypes

display(Markdown("**Schema dict:**"))
for col_name, dtype in ohlcv_pl.schema.items():
    print(f"  {col_name:<20s} {dtype}")

Schema dict

id Int64 symbol String date Date open Float64 high Float64 low Float64 close Float64 adj_close Float64 volume Int64 dividends Float64 stock_splits Float64 is_filled Boolean

display(Markdown("**dtypes list:**"))
print(ohlcv_pl.dtypes)

dtypes list

[Int64, String, Date, Float64, Float64, Float64, Float64, Float64, Int64, Float64, Float64, Boolean]

---

Value Counts / Unique / N-Unique

Understanding cardinality and frequency distribution of columns. Pandas uses .value_counts(), .nunique(), and .unique(). Polars uses .value_counts(), .n_unique(), and .unique().

Pandas | value_counts()

• Value Counts: Count occurrences of each unique value.

display(Markdown("**Top 10 tickers by row count:**"))
display(ohlcv_pd["symbol"].value_counts().head(10))

Top 10 tickers by row count

	count
symbol
ABI.BR	1331
AD.AS	1331
ADYEN.AS	1331
AI.PA	1331
AIR.PA	1331
ARGX.BR	1331
ASML.AS	1331
CS.PA	1331
DG.PA	1331
BN.PA	1331

Pandas | nunique() and unique()

display(Markdown("**Number of unique values per column:**"))
display(ohlcv_pd.nunique())

Number of unique values per column

	0
id	66355
symbol	50
date	1331
open	29671
high	31651
low	31695
close	31505
adj_close	57739
volume	65199
dividends	216
stock_splits	6
is_filled	2

display(Markdown("**Unique tickers (first 10):**"))
print(ohlcv_pd["symbol"].unique()[:10])

Unique tickers (first 10)

[‘ABI.BR’ ‘AD.AS’ ‘ADS.DE’ ‘ADYEN.AS’ ‘AI.PA’ ‘AIR.PA’ ‘ALV.DE’ ‘ARGX.BR’ ‘ASML.AS’ ‘BAS.DE’]

Polars | value_counts()

• Value Counts: Count occurrences of each unique value.

display(Markdown("**Top 10 tickers by row count:**"))
display(
    ohlcv_pl.get_column("symbol")
    .value_counts()
    .sort("count", descending=True)
    .head(10)
)

Top 10 tickers by row count

symbol	count
str	u32
SAN.PA	1331
ADYEN.AS	1331
PRX.AS	1331
ARGX.BR	1331
BN.PA	1331
DSY.PA	1331
BNP.PA	1331
TTE.PA	1331
ASML.AS	1331
CS.PA	1331

Polars | n_unique() and unique()

• N Unique: Count the number of distinct values.

display(Markdown("**n_unique per column:**"))
display(
    ohlcv_pl.select(pl.all().n_unique())
)

n_unique per column

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32
66355	50	1331	29671	31651	31695	31505	57739	65199	216	6	2

display(Markdown("**Unique tickers (first 10):**"))
print(ohlcv_pl.get_column("symbol").unique().sort().head(10).to_list())

Unique tickers (first 10)

[‘ABI.BR’, ‘AD.AS’, ‘ADS.DE’, ‘ADYEN.AS’, ‘AI.PA’, ‘AIR.PA’, ‘ALV.DE’, ‘ARGX.BR’, ‘ASML.AS’, ‘BAS.DE’]

---

Null / Missing Value Inspection

Null detection is critical for data quality. Pandas uses NaN (float) for missing values, so .isna() and .isnull() are equivalent. Polars uses native Arrow null — use .is_null() and .null_count().

Pandas vs Polars | Null representation

Pandas .isna() detects both NaN and None. Polars .is_null() detects only Arrow null — NaN is a valid float value in Polars, not a null. Use .is_nan() to detect NaN specifically in Polars float columns.

Pandas | isna() / isnull()

display(Markdown("**Null counts per column:**"))
display(scores_pd.isnull().sum())

Null counts per column

	0
id	0
_index	0
symbol	0
score_date	0
sector	0
pe_zscore	3
pb_zscore	6
ev_ebitda_zscore	71
yield_zscore	35
relative_value_score	0
relative_value_rank	0
relative_strength	0
sma_50_ratio	0
sma_200_ratio	0
dist_from_52w_high	0
momentum_score	0
momentum_rank	0
implied_upside	0
recommendation_mean	14
price_falling_analysts_bullish	0
sentiment_score	0
sentiment_rank	0
composite_score	0
composite_rank	0
_scored_at	0
sma_30_close	0
sma_90_close	0
market_cap	0
index_weight	0
short_name	0
country	0
current_price	0
day_change_pct	0
five_day_change_pct	0
ytd_change_pct	0
currency	0

display(Markdown("**Null percentage per column:**"))
null_pct = (scores_pd.isnull().sum() / len(scores_pd) * 100).round(2)
display(null_pct[null_pct > 0])

Null percentage per column

	0
pe_zscore	0.64
pb_zscore	1.29
ev_ebitda_zscore	15.24
yield_zscore	7.51
recommendation_mean	3.00

Pandas | rows with any null

rows_with_nulls = scores_pd[scores_pd.isnull().any(axis=1)]
print(f"Rows with at least one null: {len(rows_with_nulls):,}")
if len(rows_with_nulls) > 0:
    display(rows_with_nulls.head())

Rows with at least one null: 120

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
0	163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.261140	NaN	2.388962	1.521163	1	0.016123	1.009090	1.130264	0.082486	0.477966	16	0.153157	1.84211	False	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085000	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.320	0.011437	-0.073156	0.105582	EUR
5	196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	NaN	False	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.600	0.013786	-0.048756	-0.090390	EUR
8	188	euro_stoxx_50	SAN.MC	2026-03-04	Financial Services	0.458599	0.499398	NaN	-1.107590	-0.049864	33	0.393197	0.953824	1.139519	0.113499	0.519307	14	0.224704	1.70000	False	0.448776	15	0.306073	9	2026-03-04 22:40:25.489180	10.604900	9.919500	145955749888	0.028570	BANCO SANTANDER S.A.	Spain	9.982	0.038818	-0.105876	-0.008739	EUR
15	176	euro_stoxx_50	ISP.MI	2026-03-04	Financial Services	0.366619	0.488302	NaN	0.723361	0.526094	13	-0.065247	0.922492	0.990855	0.119662	-0.122626	33	0.254150	2.00000	False	0.146959	21	0.183476	16	2026-03-04 22:40:25.489180	5.834933	5.769367	94268317696	0.018452	INTESA SANPAOLO	Italy	5.422	0.018216	-0.067103	-0.084276	EUR
16	195	euro_stoxx_50	UCG.MI	2026-03-04	Financial Services	0.452501	0.355225	NaN	-0.260674	0.182351	26	0.085867	0.950889	1.054430	0.137862	0.123670	22	0.261521	1.94444	False	0.240819	18	0.182280	17	2026-03-04 22:40:25.489180	73.253333	68.983556	103066501120	0.020174	UNICREDIT	Italy	68.790	0.027483	-0.072161	-0.030034	EUR

Polars | is_null() / null_count()

• Null Count: Count missing values per column.

display(Markdown("**Null counts per column (Polars):**"))
display(scores_pl.null_count())

Null counts per column (Polars)

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32
0	0	0	0	0	3	6	71	35	0	0	0	0	0	0	0	0	0	14	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0

display(Markdown("**Null percentage per column (Polars):**"))
display(
    scores_pl.select(
        (pl.all().null_count() / pl.len() * 100).round(2).name.suffix("_null_pct")
    )
)

Null percentage per column (Polars)

id_null_pct	_index_null_pct	symbol_null_pct	score_date_null_pct	sector_null_pct	pe_zscore_null_pct	pb_zscore_null_pct	ev_ebitda_zscore_null_pct	yield_zscore_null_pct	relative_value_score_null_pct	relative_value_rank_null_pct	relative_strength_null_pct	sma_50_ratio_null_pct	sma_200_ratio_null_pct	dist_from_52w_high_null_pct	momentum_score_null_pct	momentum_rank_null_pct	implied_upside_null_pct	recommendation_mean_null_pct	price_falling_analysts_bullish_null_pct	sentiment_score_null_pct	sentiment_rank_null_pct	composite_score_null_pct	composite_rank_null_pct	_scored_at_null_pct	sma_30_close_null_pct	sma_90_close_null_pct	market_cap_null_pct	index_weight_null_pct	short_name_null_pct	country_null_pct	current_price_null_pct	day_change_pct_null_pct	five_day_change_pct_null_pct	ytd_change_pct_null_pct	currency_null_pct
f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64
0.0	0.0	0.0	0.0	0.0	0.64	1.29	15.24	7.51	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	3.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

Polars | rows with any null

mask = pl.any_horizontal(pl.all().is_null())
rows_with_nulls_pl = scores_pl.filter(mask)
print(f"Rows with at least one null: {rows_with_nulls_pl.shape[0]:,}")
if rows_with_nulls_pl.shape[0] > 0:
    display(rows_with_nulls_pl.head())

Rows with at least one null: 120

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.26114	null	2.388962	1.521163	1	0.016123	1.00909	1.130264	0.082486	0.477966	16	0.153157	1.84211	false	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	null	false	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.6	0.013786	-0.048756	-0.09039	EUR
188	euro_stoxx_50	SAN.MC	2026-03-04	Financial Services	0.458599	0.499398	null	-1.10759	-0.049864	33	0.393197	0.953824	1.139519	0.113499	0.519307	14	0.224704	1.7	false	0.448776	15	0.306073	9	2026-03-04 22:40:25.489180	10.6049	9.9195	145955749888	0.02857	BANCO SANTANDER S.A.	Spain	9.982	0.038818	-0.105876	-0.008739	EUR
176	euro_stoxx_50	ISP.MI	2026-03-04	Financial Services	0.366619	0.488302	null	0.723361	0.526094	13	-0.065247	0.922492	0.990855	0.119662	-0.122626	33	0.25415	2.0	false	0.146959	21	0.183476	16	2026-03-04 22:40:25.489180	5.834933	5.769367	94268317696	0.018452	INTESA SANPAOLO	Italy	5.422	0.018216	-0.067103	-0.084276	EUR
195	euro_stoxx_50	UCG.MI	2026-03-04	Financial Services	0.452501	0.355225	null	-0.260674	0.182351	26	0.085867	0.950889	1.05443	0.137862	0.12367	22	0.261521	1.94444	false	0.240819	18	0.18228	17	2026-03-04 22:40:25.489180	73.253333	68.983556	103066501120	0.020174	UNICREDIT	Italy	68.79	0.027483	-0.072161	-0.030034	EUR

---

Exploring the Scores Dataset

The scores_daily dataset contains composite factor scores with real null values in several columns — ideal for practicing null inspection and data quality assessment.

Pandas | quick profile

display(Markdown("**scores_daily — head:**"))
display(scores_pd.head())

scores_daily — head

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
0	163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.261140	NaN	2.388962	1.521163	1	0.016123	1.009090	1.130264	0.082486	0.477966	16	0.153157	1.84211	False	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085000	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.320	0.011437	-0.073156	0.105582	EUR
1	168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.120650	1.112416	0.055524	0.685752	8	0.121212	1.33333	False	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838000	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.000	0.011649	-0.019608	0.193059	EUR
2	174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.37500	False	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.066490	0.164723	EUR
3	172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.051850	2.541889	1	0.075269	1.80000	False	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675000	129.122000	139207262208	0.027249	Siemens Energy AG	Germany	162.750	0.047297	-0.039256	0.351744	EUR
4	149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	False	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342000	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.480	-0.017073	-0.040762	0.174499	EUR

display(Markdown("**scores_daily — describe:**"))
display(scores_pd.describe(include="all"))

scores_daily — describe

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
count	466.000000	466	466	466	466	463.000000	460.000000	395.000000	431.000000	466.000000	466.000000	466.000000	466.000000	466.000000	466.000000	466.000000	466.000000	466.000000	452.000000	466	466.000000	466.000000	466.000000	466.000000	466	466.000000	466.000000	4.660000e+02	466.000000	466	466	466.000000	466.000000	466.000000	466.000000	466
unique	NaN	4	167	3	10	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	2	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	167	16	NaN	NaN	NaN	NaN	6
top	NaN	stoxx_asia_50	CVX	2026-03-12	Financial Services	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	False	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Chevron Corporation	United States	NaN	NaN	NaN	NaN	USD
freq	NaN	150	4	169	96	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	401	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	4	159	NaN	NaN	NaN	NaN	167
mean	2083.712446	NaN	NaN	NaN	NaN	0.011386	0.024513	0.038046	0.051915	0.036204	24.733906	0.047972	0.991152	1.064981	0.154695	-0.000157	24.768240	0.175049	2.011047	NaN	-0.004157	24.793991	0.010630	24.774678	2026-03-08 09:35:56.501645824	2726.980722	2568.225890	3.030703e+12	0.021271	NaN	NaN	2721.358232	-0.001651	-0.024835	0.028935	NaN
min	149.000000	NaN	NaN	NaN	NaN	-3.244355	-3.609853	-3.562197	-1.765215	-3.069151	1.000000	-0.758111	0.764715	0.630163	0.000686	-2.270682	1.000000	-0.287205	1.224490	NaN	-3.028190	1.000000	-1.265925	1.000000	2026-03-04 22:40:25.489180	4.967667	4.912000	1.516982e+10	0.000102	NaN	NaN	5.120000	-0.071643	-0.147762	-0.326764	NaN
25%	266.250000	NaN	NaN	NaN	NaN	-0.544030	-0.360416	-0.292057	-0.680445	-0.327241	12.000000	-0.207124	0.928823	0.946901	0.073301	-0.442542	12.000000	0.070549	1.666670	NaN	-0.632834	12.000000	-0.175500	12.000000	2026-03-04 22:40:25.489179904	71.671500	72.912944	9.705579e+10	0.007191	NaN	NaN	69.900000	-0.013312	-0.054816	-0.083552	NaN
50%	2387.500000	NaN	NaN	NaN	NaN	0.294467	0.328987	0.309876	0.051261	0.178100	24.000000	-0.027879	0.984705	1.058055	0.124107	0.029804	24.000000	0.152967	1.944440	NaN	0.023661	24.000000	0.031939	24.000000	2026-03-07 03:00:57.486865920	206.003000	202.193333	2.577942e+11	0.012848	NaN	NaN	203.470000	-0.002397	-0.023506	0.019151	NaN
75%	3410.750000	NaN	NaN	NaN	NaN	0.643742	0.617103	0.622976	0.718698	0.523516	37.000000	0.197873	1.046022	1.177306	0.200560	0.488433	37.000000	0.273151	2.280000	NaN	0.527927	37.000000	0.243820	37.000000	2026-03-12 12:52:26.509884928	856.826917	837.629000	1.628276e+12	0.027448	NaN	NaN	799.057500	0.009959	0.002131	0.129549	NaN
max	3527.000000	NaN	NaN	NaN	NaN	1.666177	2.019105	1.697527	2.919889	1.521163	50.000000	3.342994	1.233011	1.905060	0.589001	2.806741	50.000000	0.804817	4.785710	NaN	2.169723	50.000000	1.287144	50.000000	2026-03-12 12:52:26.509885	68837.666667	60523.222222	4.587752e+13	0.245721	NaN	NaN	69950.000000	0.091834	0.192987	0.466977	NaN
std	1340.387660	NaN	NaN	NaN	NaN	0.894679	0.922349	0.875291	0.914714	0.723838	14.457279	0.462482	0.081733	0.182469	0.117997	0.817166	14.472144	0.172802	0.512925	NaN	0.886008	14.458262	0.336210	14.470685	NaN	9736.666450	8945.961901	6.558209e+12	0.024398	NaN	NaN	9820.254826	0.022118	0.046550	0.151246	NaN

display(Markdown("**scores_daily — null counts:**"))
display(scores_pd.isnull().sum())

scores_daily — null counts

	0
id	0
_index	0
symbol	0
score_date	0
sector	0
pe_zscore	3
pb_zscore	6
ev_ebitda_zscore	71
yield_zscore	35
relative_value_score	0
relative_value_rank	0
relative_strength	0
sma_50_ratio	0
sma_200_ratio	0
dist_from_52w_high	0
momentum_score	0
momentum_rank	0
implied_upside	0
recommendation_mean	14
price_falling_analysts_bullish	0
sentiment_score	0
sentiment_rank	0
composite_score	0
composite_rank	0
_scored_at	0
sma_30_close	0
sma_90_close	0
market_cap	0
index_weight	0
short_name	0
country	0
current_price	0
day_change_pct	0
five_day_change_pct	0
ytd_change_pct	0
currency	0

Polars | quick profile

display(Markdown("**scores_daily — head (Polars):**"))
display(scores_pl.head())

scores_daily — head (Polars)

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.26114	null	2.388962	1.521163	1	0.016123	1.00909	1.130264	0.082486	0.477966	16	0.153157	1.84211	false	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.12065	1.112416	0.055524	0.685752	8	0.121212	1.33333	false	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.0	0.011649	-0.019608	0.193059	EUR
174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.375	false	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.06649	0.164723	EUR
172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.05185	2.541889	1	0.075269	1.8	false	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675	129.122	139207262208	0.027249	Siemens Energy AG	Germany	162.75	0.047297	-0.039256	0.351744	EUR
149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	false	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.48	-0.017073	-0.040762	0.174499	EUR

display(Markdown("**scores_daily — describe (Polars):**"))
display(scores_pl.describe())

scores_daily — describe (Polars)

statistic	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
str	f64	str	str	str	str	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	f64	str	f64	f64	f64	f64	str	str	f64	f64	f64	f64	str
count	466.0	466	466	466	466	463.0	460.0	395.0	431.0	466.0	466.0	466.0	466.0	466.0	466.0	466.0	466.0	466.0	452.0	466.0	466.0	466.0	466.0	466.0	466	466.0	466.0	466.0	466.0	466	466	466.0	466.0	466.0	466.0	466
null_count	0.0	0	0	0	0	3.0	6.0	71.0	35.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	14.0	0.0	0.0	0.0	0.0	0.0	0	0.0	0.0	0.0	0.0	0	0	0.0	0.0	0.0	0.0	0
mean	2083.712446	null	null	2026-03-07 20:48:24.721030	null	0.011386	0.024513	0.038046	0.051915	0.036204	24.733906	0.047972	0.991152	1.064981	0.154695	-0.000157	24.76824	0.175049	2.011047	0.139485	-0.004157	24.793991	0.01063	24.774678	2026-03-08 09:35:56.501646	2726.980722	2568.22589	3.0307e12	0.021271	null	null	2721.358232	-0.001651	-0.024835	0.028935	null
std	1340.38766	null	null	null	null	0.894679	0.922349	0.875291	0.914714	0.723838	14.457279	0.462482	0.081733	0.182469	0.117997	0.817166	14.472144	0.172802	0.512925	null	0.886008	14.458262	0.33621	14.470685	null	9736.66645	8945.961901	6.5582e12	0.024398	null	null	9820.254826	0.022118	0.04655	0.151246	null
min	149.0	euro_stoxx_50	0388.HK	2026-03-04	Basic Materials	-3.244355	-3.609853	-3.562197	-1.765215	-3.069151	1.0	-0.758111	0.764715	0.630163	0.000686	-2.270682	1.0	-0.287205	1.22449	0.0	-3.02819	1.0	-1.265925	1.0	2026-03-04 22:40:25.489180	4.967667	4.912	1.5170e10	0.000102	AB INBEV	Australia	5.12	-0.071643	-0.147762	-0.326764	AUD
25%	266.0	null	null	2026-03-04	null	-0.542352	-0.358338	-0.285053	-0.678854	-0.327398	12.0	-0.207295	0.928634	0.94669	0.073233	-0.443075	12.0	0.070336	1.66667	null	-0.635563	12.0	-0.176007	12.0	2026-03-04 22:40:25.489179	71.527	72.844222	9.7039e10	0.00719	null	null	69.8	-0.013336	-0.054849	-0.083791	null
50%	2388.0	null	null	2026-03-07	null	0.294467	0.32996	0.309876	0.051261	0.179359	24.0	-0.025968	0.984787	1.058079	0.124181	0.031184	24.0	0.153157	1.94444	null	0.024154	24.0	0.032344	24.0	2026-03-07 03:00:57.486865	207.221667	204.365333	2.5787e11	0.013111	null	null	204.83	-0.002381	-0.023153	0.019395	null
75%	3411.0	null	null	2026-03-12	null	0.645507	0.616745	0.625587	0.723361	0.526094	37.0	0.199916	1.046095	1.17734	0.200735	0.489412	37.0	0.273617	2.28	null	0.528756	37.0	0.24484	37.0	2026-03-12 12:52:26.509884	900.273667	874.914222	1.6312e12	0.027458	null	null	821.42	0.009963	0.002179	0.12961	null
max	3527.0	stoxx_usa_50	XOM	2026-03-12	Utilities	1.666177	2.019105	1.697527	2.919889	1.521163	50.0	3.342994	1.233011	1.90506	0.589001	2.806741	50.0	0.804817	4.78571	1.0	2.169723	50.0	1.287144	50.0	2026-03-12 12:52:26.509885	68837.666667	60523.222222	4.5878e13	0.245721	adidas AG	United States	69950.0	0.091834	0.192987	0.466977	USD

display(Markdown("**scores_daily — null counts (Polars):**"))
display(scores_pl.null_count())

scores_daily — null counts (Polars)

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32	u32
0	0	0	0	0	3	6	71	35	0	0	0	0	0	0	0	0	0	14	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0

---

Exploring the Dimension Table

Head Preview

display(Markdown("**index_dim — Pandas:**"))
display(dim_pd.drop(columns="long_business_summary").head())

index_dim — Pandas

	id	_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_from	valid_to	is_current
0	1	euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Materials	semiconductor-equipment-materials	Netherlands	Veldhoven	https://www.asml.com	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	2026-03-04 22:11:36.189862	None	True
1	2	euro_stoxx_50	MC.PA	LVMH Moët Hennessy - Louis Vuitton, Société Européenne	LVMH	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://www.lvmh.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.189862	None	True
2	3	euro_stoxx_50	RMS.PA	Hermès International Société en commandite par actions	HERMES INTL	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://finance.hermes.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.193940	None	True
3	4	euro_stoxx_50	OR.PA	L'Oréal S.A.	L'OREAL	Consumer Defensive	consumer-defensive	Household & Personal Products	household-personal-products	France	Clichy	https://www.loreal.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.193940	None	True
4	5	euro_stoxx_50	SAP.DE	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market	1998-04-09	2021-01-01	2026-03-04 22:11:36.193940	None	True

display(Markdown("**index_dim — Polars:**"))
display(dim_pl.head())

index_dim — Polars

id	_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	long_business_summary	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_from	valid_to	is_current
i64	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	date	date	datetime[ns]	null	bool
1	euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Mate…	semiconductor-equipment-materi…	Netherlands	Veldhoven	https://www.asml.com	ASML Holding N.V. provides lit…	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	2026-03-04 22:11:36.189862	null	true
2	euro_stoxx_50	MC.PA	LVMH Moët Hennessy - Louis Vui…	LVMH	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://www.lvmh.com	LVMH Moët Hennessy - Louis Vui…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.189862	null	true
3	euro_stoxx_50	RMS.PA	Hermès International Société e…	HERMES INTL	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://finance.hermes.com	Hermès International Société e…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.193940	null	true
4	euro_stoxx_50	OR.PA	L'Oréal S.A.	L'OREAL	Consumer Defensive	consumer-defensive	Household & Personal Products	household-personal-products	France	Clichy	https://www.loreal.com	L'Oréal S.A., through its subs…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.193940	null	true
5	euro_stoxx_50	SAP.DE	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	SAP SE, together with its subs…	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market	1998-04-09	2021-01-01	2026-03-04 22:11:36.193940	null	true

Types and Schema

display(Markdown("**index_dim — dtypes (Pandas):**"))
display(dim_pd.dtypes)

index_dim — dtypes (Pandas)

	0
id	int64
_index	object
symbol	object
long_name	object
short_name	object
sector	object
sector_key	object
industry	object
industry_key	object
country	object
city	object
website	object
long_business_summary	object
exchange	object
full_exchange_name	object
exchange_timezone_name	object
exchange_timezone_short	object
currency	object
financial_currency	object
quote_type	object
market	object
range_start	object
price_data_start	object
valid_from	datetime64[ns]
valid_to	object
is_current	bool

display(Markdown("**index_dim — schema (Polars):**"))
for col_name, dtype in dim_pl.schema.items():
    print(f"  {col_name:<25s} {dtype}")

index_dim — schema (Polars)

id Int64 _index String symbol String long_name String short_name String sector String sector_key String industry String industry_key String country String city String website String long_business_summary String exchange String full_exchange_name String exchange_timezone_name String exchange_timezone_short String currency String financial_currency String quote_type String market String range_start Date price_data_start Date valid_from Datetime(time_unit=‘ns’, time_zone=None) valid_to Null is_current Boolean

---

Data Profiling Strategies

Reusable profiling functions that combine shape, types, null counts, and basic stats into a single summary. Build these once and apply to any dataset.

A reusable profiling pattern: for each column report dtype, null count, unique count, and a few sample values.

Pandas | quick profiler

• Drop Nulls: Remove rows with missing values (Pandas).
• N Unique: Count the number of distinct values.

def profile_pd(df: pd.DataFrame) -> pd.DataFrame:
    """Return a one-row-per-column profiling DataFrame (Pandas)."""
    records = []
    for col in df.columns:
        records.append({
            "column":     col,
            "dtype":      str(df[col].dtype),
            "null_count": int(df[col].isnull().sum()),
            "null_pct":   round(df[col].isnull().mean() * 100, 2),
            "n_unique":   int(df[col].nunique()),
            "sample":     str(df[col].dropna().iloc[:3].tolist()),
        })
    return pd.DataFrame(records)
 
display(Markdown("**OHLCV profile (Pandas):**"))
display(profile_pd(ohlcv_pd))

OHLCV profile (Pandas)

	column	dtype	null_count	null_pct	n_unique	sample
0	id	int64	0	0.0	66355	[21160, 21161, 21162]
1	symbol	object	0	0.0	50	['ABI.BR', 'ABI.BR', 'ABI.BR']
2	date	datetime64[ns]	0	0.0	1331	[Timestamp('2021-01-04 00:00:00'), Timestamp('2021-01-05 00:00:00'), Timestamp('2021-01-06 00:00:00')]
3	open	float64	0	0.0	29671	[58.15, 56.9, 57.96]
4	high	float64	0	0.0	31651	[58.85, 57.98, 58.94]
5	low	float64	0	0.0	31695	[56.78, 56.75, 57.39]
6	close	float64	0	0.0	31505	[57.21, 57.18, 58.77]
7	adj_close	float64	0	0.0	57739	[53.5761, 53.548, 55.037]
8	volume	int64	0	0.0	65199	[1513937, 1382722, 1370204]
9	dividends	float64	0	0.0	216	[0.0, 0.0, 0.0]
10	stock_splits	float64	0	0.0	6	[0.0, 0.0, 0.0]
11	is_filled	bool	0	0.0	2	[False, False, False]

Polars | quick profiler

• Null Count: Count missing values per column.
• Drop Nulls: Remove rows with missing values (Polars).
• N Unique: Count the number of distinct values.

def profile_pl(df: pl.DataFrame) -> pl.DataFrame:
    """Return a one-row-per-column profiling DataFrame (Polars)."""
    rows = []
    for col_name in df.columns:
        col = df.get_column(col_name)
        rows.append({
            "column":     col_name,
            "dtype":      str(col.dtype),
            "null_count": col.null_count(),
            "null_pct":   round(col.null_count() / df.height * 100, 2),
            "n_unique":   col.n_unique(),
            "sample":     str(col.drop_nulls().head(3).to_list()),
        })
    return pl.DataFrame(rows)
 
display(Markdown("**OHLCV profile (Polars):**"))
display(profile_pl(ohlcv_pl))

OHLCV profile (Polars)

column	dtype	null_count	null_pct	n_unique	sample
str	str	i64	f64	i64	str
id	Int64	0	0.0	66355	[21160, 21161, 21162]
symbol	String	0	0.0	50	['ABI.BR', 'ABI.BR', 'ABI.BR']
date	Date	0	0.0	1331	[datetime.date(2021, 1, 4), da…
open	Float64	0	0.0	29671	[58.15, 56.9, 57.96]
high	Float64	0	0.0	31651	[58.85, 57.98, 58.94]
low	Float64	0	0.0	31695	[56.78, 56.75, 57.39]
close	Float64	0	0.0	31505	[57.21, 57.18, 58.77]
adj_close	Float64	0	0.0	57739	[53.5761, 53.548, 55.037]
volume	Int64	0	0.0	65199	[1513937, 1382722, 1370204]
dividends	Float64	0	0.0	216	[0.0, 0.0, 0.0]
stock_splits	Float64	0	0.0	6	[0.0, 0.0, 0.0]
is_filled	Boolean	0	0.0	2	[False, False, False]

Profile all three datasets

Profile eurostoxx50_ohlcv — Pandas vs Polars describe comparison

prof_pd = profile_pd(ohlcv_pd)
display(Markdown("*Pandas profile:*"))
display(prof_pd)

Pandas profile:

	column	dtype	null_count	null_pct	n_unique	sample
0	id	int64	0	0.0	66355	[21160, 21161, 21162]
1	symbol	object	0	0.0	50	['ABI.BR', 'ABI.BR', 'ABI.BR']
2	date	object	0	0.0	1331	[datetime.date(2021, 1, 4), datetime.date(2021, 1, 5), datetime.date(2021, 1, 6)]
3	open	float64	0	0.0	29671	[58.15, 56.9, 57.96]
4	high	float64	0	0.0	31651	[58.85, 57.98, 58.94]
5	low	float64	0	0.0	31695	[56.78, 56.75, 57.39]
6	close	float64	0	0.0	31505	[57.21, 57.18, 58.77]
7	adj_close	float64	0	0.0	57739	[53.5761, 53.548, 55.037]
8	volume	int64	0	0.0	65199	[1513937, 1382722, 1370204]
9	dividends	float64	0	0.0	216	[0.0, 0.0, 0.0]
10	stock_splits	float64	0	0.0	6	[0.0, 0.0, 0.0]
11	is_filled	bool	0	0.0	2	[False, False, False]

prof_pl = profile_pl(ohlcv_pl)
display(Markdown("*Polars profile:*"))
display(prof_pl)

Polars profile:

column	dtype	null_count	null_pct	n_unique	sample
str	str	i64	f64	i64	str
id	Int64	0	0.0	66355	[21160, 21161, 21162]
symbol	String	0	0.0	50	['ABI.BR', 'ABI.BR', 'ABI.BR']
date	Date	0	0.0	1331	[datetime.date(2021, 1, 4), da…
open	Float64	0	0.0	29671	[58.15, 56.9, 57.96]
high	Float64	0	0.0	31651	[58.85, 57.98, 58.94]
low	Float64	0	0.0	31695	[56.78, 56.75, 57.39]
close	Float64	0	0.0	31505	[57.21, 57.18, 58.77]
adj_close	Float64	0	0.0	57739	[53.5761, 53.548, 55.037]
volume	Int64	0	0.0	65199	[1513937, 1382722, 1370204]
dividends	Float64	0	0.0	216	[0.0, 0.0, 0.0]
stock_splits	Float64	0	0.0	6	[0.0, 0.0, 0.0]
is_filled	Boolean	0	0.0	2	[False, False, False]

Profile index_dim — Pandas vs Polars describe comparison

prof_pd = profile_pd(dim_pd)
prof_pd = prof_pd[prof_pd["column"] != "long_business_summary"]
display(Markdown("*Pandas profile:*"))
display(prof_pd.head())

Pandas profile:

	column	dtype	null_count	null_pct	n_unique	sample
0	id	int64	0	0.0	169	[1, 2, 3]
1	_index	object	0	0.0	4	['euro_stoxx_50', 'euro_stoxx_50', 'euro_stoxx_50']
2	symbol	object	0	0.0	167	['ASML.AS', 'MC.PA', 'RMS.PA']
3	long_name	object	0	0.0	166	['ASML Holding N.V.', 'LVMH Moët Hennessy - Louis Vuitton, Société Européenne', 'Hermès International Société en commandite par actions']
4	short_name	object	0	0.0	167	['ASML HOLDING', 'LVMH', 'HERMES INTL']

prof_pl = profile_pl(dim_pl)
prof_pl = prof_pl.filter(pl.col("column") != "long_business_summary")
display(Markdown("*Polars profile:*"))
display(prof_pl.head())

Polars profile:

column	dtype	null_count	null_pct	n_unique	sample
str	str	i64	f64	i64	str
id	Int64	0	0.0	169	[1, 2, 3]
_index	String	0	0.0	4	['euro_stoxx_50', 'euro_stoxx_…
symbol	String	0	0.0	167	['ASML.AS', 'MC.PA', 'RMS.PA']
long_name	String	0	0.0	166	['ASML Holding N.V.', 'LVMH Mo…
short_name	String	0	0.0	167	['ASML HOLDING', 'LVMH', 'HERM…

Profile scores_daily — Pandas vs Polars describe comparison

prof_pd = profile_pd(scores_pd)
display(Markdown("*Pandas profile:*"))
display(prof_pd.head())

Pandas profile:

	column	dtype	null_count	null_pct	n_unique	sample
0	id	int64	0	0.0	466	[163, 168, 174]
1	_index	object	0	0.0	4	['euro_stoxx_50', 'euro_stoxx_50', 'euro_stoxx_50']
2	symbol	object	0	0.0	167	['BNP.PA', 'DTE.DE', 'IFX.DE']
3	score_date	object	0	0.0	3	[datetime.date(2026, 3, 4), datetime.date(2026, 3, 4), datetime.date(2026, 3, 4)]
4	sector	object	0	0.0	10	['Financial Services', 'Communication Services', 'Technology']

prof_pl = profile_pl(scores_pl)
display(Markdown("*Polars profile:*"))
display(prof_pl.head())

Polars profile:

column	dtype	null_count	null_pct	n_unique	sample
str	str	i64	f64	i64	str
id	Int64	0	0.0	466	[163, 168, 174]
_index	String	0	0.0	4	['euro_stoxx_50', 'euro_stoxx_…
symbol	String	0	0.0	167	['BNP.PA', 'DTE.DE', 'IFX.DE']
score_date	Date	0	0.0	3	[datetime.date(2026, 3, 4), da…
sector	String	0	0.0	10	['Financial Services', 'Commun…

---

Comparison Table | Pandas vs Polars

comparison = """
| Task | Pandas | Polars |
|---|---|---|
| First N rows | `df.head(n)` | `df.head(n)` |
| Last N rows | `df.tail(n)` | `df.tail(n)` |
| Random sample | `df.sample(n)` | `df.sample(n)` |
| Transposed preview | (no built-in) | `df.glimpse()` |
| Shape | `df.shape` | `df.shape` |
| Describe (numeric) | `df.describe()` | `df.describe()` |
| Describe (all) | `df.describe(include="all")` | `df.describe()` (all by default) |
| Column dtypes | `df.dtypes` | `df.dtypes` / `df.schema` |
| Info summary | `df.info()` | (no direct equivalent) |
| Value counts | `s.value_counts()` | `s.value_counts()` |
| Unique values | `s.unique()` | `s.unique()` |
| N-unique (one col) | `s.nunique()` | `s.n_unique()` |
| N-unique (all cols) | `df.nunique()` | `df.select(pl.all().n_unique())` |
| Null count (col) | `s.isnull().sum()` | `s.null_count()` |
| Null count (all) | `df.isnull().sum()` | `df.null_count()` |
| Null percentage | `df.isnull().mean() * 100` | `pl.all().null_count() / pl.len() * 100` |
| Filter null rows | `df[df.isnull().any(axis=1)]` | `df.filter(pl.any_horizontal(pl.all().is_null()))` |
"""
display(Markdown(comparison))

Task	Pandas	Polars
First N rows	df.head(n)	df.head(n)
Last N rows	df.tail(n)	df.tail(n)
Random sample	df.sample(n)	df.sample(n)
Transposed preview	(no built-in)	df.glimpse()
Shape	df.shape	df.shape
Describe (numeric)	df.describe()	df.describe()
Describe (all)	df.describe(include="all")	df.describe() (all by default)
Column dtypes	df.dtypes	df.dtypes / df.schema
Info summary	df.info()	(no direct equivalent)
Value counts	s.value_counts()	s.value_counts()
Unique values	s.unique()	s.unique()
N-unique (one col)	s.nunique()	s.n_unique()
N-unique (all cols)	df.nunique()	df.select(pl.all().n_unique())
Null count (col)	s.isnull().sum()	s.null_count()
Null count (all)	df.isnull().sum()	df.null_count()
Null percentage	df.isnull().mean() * 100	pl.all().null_count() / pl.len() * 100
Filter null rows	df[df.isnull().any(axis=1)]	df.filter(pl.any_horizontal(pl.all().is_null()))

---

Selecting Rows & Columns

This section covers positional, label-based, and name-based row and column selection. Pandas uses .iloc[] (positional) and .loc[] (label-based). Polars uses .select(), .filter(), and pl.col() expressions.

Pandas vs Polars | Selection philosophy

Pandas provides two indexing axes: .iloc[] for integer position and .loc[] for label-based access. Polars has no .iloc/.loc — all column selection goes through .select() with expressions, and all row filtering goes through .filter(). This eliminates the SettingWithCopyWarning and chained-indexing bugs common in Pandas.

Dataset Overview

display(Markdown("**Quick look at both datasets:**"))
display(ohlcv_pd.head(3))
display(dim_pd.drop(columns="long_business_summary").head(3))

Quick look at both datasets

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False

	id	_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_from	valid_to	is_current
0	1	euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Materials	semiconductor-equipment-materials	Netherlands	Veldhoven	https://www.asml.com	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	2026-03-04 22:11:36.189862	None	True
1	2	euro_stoxx_50	MC.PA	LVMH Moët Hennessy - Louis Vuitton, Société Européenne	LVMH	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://www.lvmh.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.189862	None	True
2	3	euro_stoxx_50	RMS.PA	Hermès International Société en commandite par actions	HERMES INTL	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://finance.hermes.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	2026-03-04 22:11:36.193940	None	True

Selecting Rows by Position

Single Row

# Pandas — iloc returns a Series
display(Markdown("**Pandas — single row as Series:**"))
display(ohlcv_pd.iloc[0])

Pandas | single row as Series

	0
id	21160
symbol	ABI.BR
date	2021-01-04
open	58.15
high	58.85
low	56.78
close	57.21
adj_close	53.5761
volume	1513937
dividends	0.0
stock_splits	0.0
is_filled	False

# Polars — row() returns a tuple, slice() returns a 1-row DataFrame
display(Markdown("**Polars — single row as tuple:**"))
print(ohlcv_pl.row(0))
 
display(Markdown("**Polars — single row as DataFrame:**"))
display(ohlcv_pl.slice(0, 1))

Polars | single row as tuple

(21160, ‘ABI.BR’, datetime.date(2021, 1, 4), 58.15, 58.85, 56.78, 57.21, 53.5761, 1513937, 0.0, 0.0, False)

Polars | single row as DataFrame

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false

Multiple Rows by Position

# Pandas — pass a list of positions
display(Markdown("**Pandas — rows at positions 0, 10, 100:**"))
display(ohlcv_pd.iloc[[0, 10, 100]])

Pandas | rows at positions 0, 10, 100

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
10	21170	ABI.BR	2021-01-18	56.25	57.30	56.20	57.08	53.4544	730298	0.0	0.0	False
100	21260	ABI.BR	2021-05-26	61.99	62.39	61.83	62.12	58.6701	940186	0.0	0.0	False

# Polars — bracket indexing with a list
display(Markdown("**Polars — rows at positions 0, 10, 100:**"))
display(ohlcv_pl[[0, 10, 100]])

Polars | rows at positions 0, 10, 100

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21170	ABI.BR	2021-01-18	56.25	57.3	56.2	57.08	53.4544	730298	0.0	0.0	false
21260	ABI.BR	2021-05-26	61.99	62.39	61.83	62.12	58.6701	940186	0.0	0.0	false

Row Slicing

# Pandas — standard Python slicing (start:stop)
display(Markdown("**Pandas — rows 10 to 14:**"))
display(ohlcv_pd.iloc[10:15])

Pandas | rows 10 to 14

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
10	21170	ABI.BR	2021-01-18	56.25	57.30	56.20	57.08	53.4544	730298	0.0	0.0	False
11	21171	ABI.BR	2021-01-19	57.10	57.26	56.26	56.35	52.7707	1116570	0.0	0.0	False
12	21172	ABI.BR	2021-01-20	56.35	56.77	56.00	56.24	52.6677	1226516	0.0	0.0	False
13	21173	ABI.BR	2021-01-21	56.20	56.55	55.31	55.31	51.7968	1404283	0.0	0.0	False
14	21174	ABI.BR	2021-01-22	55.28	55.28	54.12	54.78	51.3005	1557287	0.0	0.0	False

# Polars — slice(offset, length)
display(Markdown("**Polars — rows 10 to 14:**"))
display(ohlcv_pl.slice(10, 5))

Polars | rows 10 to 14

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21170	ABI.BR	2021-01-18	56.25	57.3	56.2	57.08	53.4544	730298	0.0	0.0	false
21171	ABI.BR	2021-01-19	57.1	57.26	56.26	56.35	52.7707	1116570	0.0	0.0	false
21172	ABI.BR	2021-01-20	56.35	56.77	56.0	56.24	52.6677	1226516	0.0	0.0	false
21173	ABI.BR	2021-01-21	56.2	56.55	55.31	55.31	51.7968	1404283	0.0	0.0	false
21174	ABI.BR	2021-01-22	55.28	55.28	54.12	54.78	51.3005	1557287	0.0	0.0	false

Row by Label

Pandas DataFrames have a row index that supports label-based access via loc. Polars has no row index — use filter() instead.

# Pandas — loc with label-based indexing
# Set 'symbol' as index to demonstrate label access
dim_indexed = dim_pd.drop(columns="long_business_summary").set_index("symbol")
display(Markdown("**Pandas — loc with label index:**"))
display(dim_indexed.loc[["ASML.AS", "SAP.DE"]])

Pandas | loc with label index

	id	_index	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_from	valid_to	is_current
symbol
ASML.AS	1	euro_stoxx_50	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Materials	semiconductor-equipment-materials	Netherlands	Veldhoven	https://www.asml.com	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	2026-03-04 22:11:36.189862	None	True
SAP.DE	5	euro_stoxx_50	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market	1998-04-09	2021-01-01	2026-03-04 22:11:36.193940	None	True

# Polars — no index, use filter instead
display(Markdown("**Polars — filter as label equivalent:**"))
display(dim_pl.filter(pl.col("symbol").is_in(["ASML.AS", "SAP.DE"])).drop("long_business_summary"))

Polars | filter as label equivalent

id	_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_from	valid_to	is_current
i64	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	date	date	datetime[ns]	null	bool
1	euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Mate…	semiconductor-equipment-materi…	Netherlands	Veldhoven	https://www.asml.com	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	2026-03-04 22:11:36.189862	null	true
5	euro_stoxx_50	SAP.DE	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market	1998-04-09	2021-01-01	2026-03-04 22:11:36.193940	null	true

Selecting Rows and Columns Together

By Position

# Pandas — iloc[rows, cols]
display(Markdown("**Pandas — rows 0-4, columns 2-5:**"))
display(ohlcv_pd.iloc[:5, 2:6])

Pandas | rows 0-4, columns 2-5

	date	open	high	low
0	2021-01-04	58.15	58.85	56.78
1	2021-01-05	56.90	57.98	56.75
2	2021-01-06	57.96	58.94	57.39
3	2021-01-07	58.68	58.86	57.88
4	2021-01-08	58.16	58.40	57.43

# Polars — slice + select by column names (no positional column indexing)
display(Markdown("**Polars — rows 0-4, columns date through close:**"))
cols = ohlcv_pl.columns[2:6]
display(ohlcv_pl.slice(0, 5).select(cols))

Polars | rows 0-4, columns date through close

date	open	high	low
date	f64	f64	f64
2021-01-04	58.15	58.85	56.78
2021-01-05	56.9	57.98	56.75
2021-01-06	57.96	58.94	57.39
2021-01-07	58.68	58.86	57.88
2021-01-08	58.16	58.4	57.43

By Name and Condition

# Pandas — loc with condition + column names
display(Markdown("**Pandas — ASML rows, selected columns:**"))
display(ohlcv_pd.loc[ohlcv_pd["symbol"] == "ASML.AS", ["date", "close", "volume"]].head())

Pandas | ASML rows, selected columns

	date	close	volume
10634	2021-01-04	406.25	789502
10635	2021-01-05	406.90	798787
10636	2021-01-06	402.85	875711
10637	2021-01-07	403.90	874780
10638	2021-01-08	416.05	975243

# Polars — filter + select
display(Markdown("**Polars — ASML rows, selected columns:**"))
display(
    ohlcv_pl
    .filter(pl.col("symbol") == "ASML.AS")
    .select("date", "close", "volume")
    .head()
)

Polars | ASML rows, selected columns

date	close	volume
date	f64	i64
2021-01-04	406.25	789502
2021-01-05	406.9	798787
2021-01-06	402.85	875711
2021-01-07	403.9	874780
2021-01-08	416.05	975243

Practical Subset from Dimension Table

# Pandas — first 5 stocks, just name and sector
display(Markdown("**Pandas:**"))
display(dim_pd.iloc[:5][["symbol", "long_name", "sector", "country"]])

Pandas | first 5 stocks — symbol, name, sector, country

	symbol	long_name	sector	country
0	ASML.AS	ASML Holding N.V.	Technology	Netherlands
1	MC.PA	LVMH Moët Hennessy - Louis Vuitton, Société Européenne	Consumer Cyclical	France
2	RMS.PA	Hermès International Société en commandite par actions	Consumer Cyclical	France
3	OR.PA	L'Oréal S.A.	Consumer Defensive	France
4	SAP.DE	SAP SE	Technology	Germany

# Polars — first 5 stocks, just name and sector
display(Markdown("**Polars:**"))
display(dim_pl.slice(0, 5).select("symbol", "long_name", "sector", "country"))

Polars | first 5 stocks — symbol, name, sector, country

symbol	long_name	sector	country
str	str	str	str
ASML.AS	ASML Holding N.V.	Technology	Netherlands
MC.PA	LVMH Moët Hennessy - Louis Vui…	Consumer Cyclical	France
RMS.PA	Hermès International Société e…	Consumer Cyclical	France
OR.PA	L'Oréal S.A.	Consumer Defensive	France
SAP.DE	SAP SE	Technology	Germany

---

Single Column Selection

Pandas | bracket and dot notation

Selects the close column from ohlcv_pd using bracket notation and dot attribute access, both returning the first 5 rows as an identical Series — demonstrating that df["col"] and df.col are interchangeable for column retrieval.

display(ohlcv_pd["close"].head())

	close
0	57.21
1	57.18
2	58.77
3	58.40
4	57.86

display(ohlcv_pd.close.head())

	close
0	57.21
1	57.18
2	58.77
3	58.40
4	57.86

Polars | select() and pl.col()

Selects symbol, date, and close from ohlcv_pl using string shorthand in select(), then repeats with explicit pl.col() expressions — confirming that bare strings and pl.col() are interchangeable column references, both returning a 5-row, 2–3 column result.

display(ohlcv_pl.select("symbol", "date", "close").head())

symbol	date	close
str	date	f64
ABI.BR	2021-01-04	57.21
ABI.BR	2021-01-05	57.18
ABI.BR	2021-01-06	58.77
ABI.BR	2021-01-07	58.4
ABI.BR	2021-01-08	57.86

display(ohlcv_pl.select(pl.col("symbol"), pl.col("close")).head())

symbol	close
str	f64
ABI.BR	57.21
ABI.BR	57.18
ABI.BR	58.77
ABI.BR	58.4
ABI.BR	57.86

---

Multiple Column Selection

Pandas | list, loc

Selects three columns from ohlcv_pd using a double-bracket list — the standard Pandas idiom for returning a DataFrame (not a Series) with a named column subset — then demonstrates loc[:, [...]] and loc[:, "open":"close"] for label-based multi-column access.

display(ohlcv_pd[["symbol", "date", "close"]].head())

	symbol	date	close
0	ABI.BR	2021-01-04	57.21
1	ABI.BR	2021-01-05	57.18
2	ABI.BR	2021-01-06	58.77
3	ABI.BR	2021-01-07	58.40
4	ABI.BR	2021-01-08	57.86

display(Markdown("**Select specific columns with `loc`:**"))
display(ohlcv_pd.loc[:, ["symbol", "open", "close"]].head())

Select specific columns with loc

	symbol	open	close
0	ABI.BR	58.15	57.21
1	ABI.BR	56.90	57.18
2	ABI.BR	57.96	58.77
3	ABI.BR	58.68	58.40
4	ABI.BR	58.16	57.86

display(Markdown("**Slice columns with `loc` (label range):**"))
display(ohlcv_pd.loc[:, "open":"close"].head())

Slice columns with loc (label range)

	open	high	low	close
0	58.15	58.85	56.78	57.21
1	56.90	57.98	56.75	57.18
2	57.96	58.94	57.39	58.77
3	58.68	58.86	57.88	58.40
4	58.16	58.40	57.43	57.86

Polars | pl.col() with a list

Passes a Python list of column names to pl.col() inside select(), returning the five OHLCV price columns — demonstrating that pl.col(["a", "b", ...]) is a concise alternative to pl.col("a"), pl.col("b"), ... for multi-column selection.

display(ohlcv_pl.select(pl.col(["symbol", "open", "high", "low", "close"])).head())

symbol	open	high	low	close
str	f64	f64	f64	f64
ABI.BR	58.15	58.85	56.78	57.21
ABI.BR	56.9	57.98	56.75	57.18
ABI.BR	57.96	58.94	57.39	58.77
ABI.BR	58.68	58.86	57.88	58.4
ABI.BR	58.16	58.4	57.43	57.86

---

Column Selection by Position

Pandas | iloc

Uses iloc[:, :3] to return the first three columns (id, symbol, date) of ohlcv_pd by position, then iloc[:, [0, 2, 4]] to retrieve non-contiguous columns id, date, and high — both operations across all rows.

display(Markdown("**First three columns by position:**"))
display(ohlcv_pd.iloc[:, :3].head())

First three columns by position

	id	symbol	date
0	21160	ABI.BR	2021-01-04
1	21161	ABI.BR	2021-01-05
2	21162	ABI.BR	2021-01-06
3	21163	ABI.BR	2021-01-07
4	21164	ABI.BR	2021-01-08

display(Markdown("**Columns at positions 0, 2, 4:**"))
display(ohlcv_pd.iloc[:, [0, 2, 4]].head())

Columns at positions 0, 2, 4

	id	date	high
0	21160	2021-01-04	58.85
1	21161	2021-01-05	57.98
2	21162	2021-01-06	58.94
3	21163	2021-01-07	58.86
4	21164	2021-01-08	58.40

Polars | index into columns list

Slices ohlcv_pl.columns (a Python list) to get the first three column names, then passes them to select() — and repeats with a list comprehension for non-contiguous positions 0, 2, 4 — since Polars has no iloc-style positional column indexer.

# Polars has no positional column indexing — slice the columns list
display(Markdown("**First three columns by position:**"))
display(ohlcv_pl.select(ohlcv_pl.columns[:3]).head())
 
display(Markdown("**Columns at positions 0, 2, 4:**"))
display(ohlcv_pl.select([ohlcv_pl.columns[i] for i in [0, 2, 4]]).head())

First three columns by position

id	symbol	date
i64	str	date
21160	ABI.BR	2021-01-04
21161	ABI.BR	2021-01-05
21162	ABI.BR	2021-01-06
21163	ABI.BR	2021-01-07
21164	ABI.BR	2021-01-08

Columns at positions 0, 2, 4

id	date	high
i64	date	f64
21160	2021-01-04	58.85
21161	2021-01-05	57.98
21162	2021-01-06	58.94
21163	2021-01-07	58.86
21164	2021-01-08	58.4

---

Select All / Exclude Columns

Pandas selects all columns by default; exclusion uses drop() (covered later). Polars provides pl.all() and pl.exclude() as expression-level selectors.

Polars | pl.all() and pl.exclude()

Uses pl.all() inside select() to pass through all 12 columns of ohlcv_pl unmodified, then pl.exclude("volume") to drop a single column and pl.exclude(["volume", "symbol"]) to drop two — demonstrating expression-level exclusion without listing every kept column.

display(ohlcv_pl.select(pl.all()).head(3))

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

display(Markdown("**All columns except volume:**"))
display(ohlcv_pl.select(pl.exclude("volume")).head())

All columns except volume

id	symbol	date	open	high	low	close	adj_close	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	0.0	0.0	false

display(Markdown("**Exclude multiple columns:**"))
display(ohlcv_pl.select(pl.exclude(["volume", "symbol"])).head())

Exclude multiple columns

id	date	open	high	low	close	adj_close	dividends	stock_splits	is_filled
i64	date	f64	f64	f64	f64	f64	f64	f64	bool
21160	2021-01-04	58.15	58.85	56.78	57.21	53.5761	0.0	0.0	false
21161	2021-01-05	56.9	57.98	56.75	57.18	53.548	0.0	0.0	false
21162	2021-01-06	57.96	58.94	57.39	58.77	55.037	0.0	0.0	false
21163	2021-01-07	58.68	58.86	57.88	58.4	54.6905	0.0	0.0	false
21164	2021-01-08	58.16	58.4	57.43	57.86	54.1848	0.0	0.0	false

---

Column Selection by Dtype

Pandas | select_dtypes()

Passes include="number" to select_dtypes() on ohlcv_pd, returning the 9 numeric columns (id, open, high, low, close, adj_close, volume, dividends, stock_splits) and filtering out symbol, date, and is_filled — then repeats with include="object" to isolate the two string columns.

display(Markdown("**Numeric columns only:**"))
display(ohlcv_pd.select_dtypes(include="number").head())

Numeric columns only

	id	open	high	low	close	adj_close	volume	dividends	stock_splits
0	21160	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0
1	21161	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0
2	21162	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0
3	21163	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0
4	21164	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0

display(Markdown("**Object / string columns only:**"))
display(ohlcv_pd.select_dtypes(include="object").head())

Object / string columns only

	symbol	date
0	ABI.BR	2021-01-04
1	ABI.BR	2021-01-05
2	ABI.BR	2021-01-06
3	ABI.BR	2021-01-07
4	ABI.BR	2021-01-08

Polars | polars.selectors

Uses cs.numeric(), cs.string(), and cs.temporal() on ohlcv_pl to select 9 numeric columns, 1 string column (symbol), and 1 temporal column (date) respectively — then demonstrates cs.by_dtype(pl.Float64) and cs.by_dtype(pl.Int64) for single-dtype filtering.

display(ohlcv_pl.select(cs.numeric()).head())

id	open	high	low	close	adj_close	volume	dividends	stock_splits
i64	f64	f64	f64	f64	f64	i64	f64	f64
21160	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0
21161	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0
21162	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0
21163	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0
21164	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0

display(ohlcv_pl.select(cs.string()).head())

symbol
str
ABI.BR
ABI.BR
ABI.BR
ABI.BR
ABI.BR

display(ohlcv_pl.select(cs.temporal()).head())

date
date
2021-01-04
2021-01-05
2021-01-06
2021-01-07
2021-01-08

display(Markdown("**Float64 columns only:**"))
display(ohlcv_pl.select(cs.by_dtype(pl.Float64)).head())

Float64 columns only

open	high	low	close	adj_close	dividends	stock_splits
f64	f64	f64	f64	f64	f64	f64
58.15	58.85	56.78	57.21	53.5761	0.0	0.0
56.9	57.98	56.75	57.18	53.548	0.0	0.0
57.96	58.94	57.39	58.77	55.037	0.0	0.0
58.68	58.86	57.88	58.4	54.6905	0.0	0.0
58.16	58.4	57.43	57.86	54.1848	0.0	0.0

display(Markdown("**Int64 columns only:**"))
display(ohlcv_pl.select(cs.by_dtype(pl.Int64)).head())

Int64 columns only

id	volume
i64	i64
21160	1513937
21161	1382722
21162	1370204
21163	1469911
21164	1428681

---

Column Selection by Pattern / Regex

Pandas | filter(regex=…)

Applies df.filter(regex="o") to ohlcv_pd, returning the 7 columns whose names contain the letter “o” (symbol, open, low, close, adj_close, volume, stock_splits) — then demonstrates anchored patterns: ^c for names starting with “c” and e for names containing “e”.

display(Markdown("**Columns matching regex (contains 'o'):**"))
display(ohlcv_pd.filter(regex="o").head())

Columns matching regex (contains ‘o’)

	symbol	open	low	close	adj_close	volume	stock_splits
0	ABI.BR	58.15	56.78	57.21	53.5761	1513937	0.0
1	ABI.BR	56.90	56.75	57.18	53.5480	1382722	0.0
2	ABI.BR	57.96	57.39	58.77	55.0370	1370204	0.0
3	ABI.BR	58.68	57.88	58.40	54.6905	1469911	0.0
4	ABI.BR	58.16	57.43	57.86	54.1848	1428681	0.0

display(Markdown("**Columns starting with 'c':**"))
display(ohlcv_pd.filter(regex="^c").head())

Columns starting with ‘c’

	close
0	57.21
1	57.18
2	58.77
3	58.40
4	57.86

display(Markdown("**Columns whose name contains 'e':**"))
display(ohlcv_pd.filter(regex="e").head())

Columns whose name contains ‘e’

	date	open	close	adj_close	volume	dividends	is_filled
0	2021-01-04	58.15	57.21	53.5761	1513937	0.0	False
1	2021-01-05	56.90	57.18	53.5480	1382722	0.0	False
2	2021-01-06	57.96	58.77	55.0370	1370204	0.0	False
3	2021-01-07	58.68	58.40	54.6905	1469911	0.0	False
4	2021-01-08	58.16	57.86	54.1848	1428681	0.0	False

Polars | pl.col(“^regex$”) and cs.by_name()

Uses cs.by_name("open", "close") to select two columns by exact name, then demonstrates Polars regex column selection with pl.col("^(c|o).*$") for names starting with “c” or “o” and pl.col("^.*e$") for names ending with “e” — all anchored with ^...$ as Polars requires.

display(ohlcv_pl.select(cs.by_name("open", "close")).head())

open	close
f64	f64
58.15	57.21
56.9	57.18
57.96	58.77
58.68	58.4
58.16	57.86

display(Markdown("**Columns whose name starts with 'c' or 'o' (regex):**"))
display(ohlcv_pl.select(pl.col("^(c|o).*$")).head())

Columns whose name starts with ‘c’ or ‘o’ (regex)

open	close
f64	f64
58.15	57.21
56.9	57.18
57.96	58.77
58.68	58.4
58.16	57.86

display(Markdown("**Columns ending with 'e':**"))
display(ohlcv_pl.select(pl.col("^.*e$")).head())

Columns ending with ‘e’

date	close	adj_close	volume
date	f64	f64	i64
2021-01-04	57.21	53.5761	1513937
2021-01-05	57.18	53.548	1382722
2021-01-06	58.77	55.037	1370204
2021-01-07	58.4	54.6905	1469911
2021-01-08	57.86	54.1848	1428681

---

Combining Selectors (Polars)

polars.selectors supports set operations: | (union), - (difference), ~ (invert).

display(Markdown("**Numeric BUT NOT Int64:**"))
display(ohlcv_pl.select(cs.numeric() - cs.by_dtype(pl.Int64)).head())

Numeric BUT NOT Int64

open	high	low	close	adj_close	dividends	stock_splits
f64	f64	f64	f64	f64	f64	f64
58.15	58.85	56.78	57.21	53.5761	0.0	0.0
56.9	57.98	56.75	57.18	53.548	0.0	0.0
57.96	58.94	57.39	58.77	55.037	0.0	0.0
58.68	58.86	57.88	58.4	54.6905	0.0	0.0
58.16	58.4	57.43	57.86	54.1848	0.0	0.0

display(Markdown("**Numeric OR temporal:**"))
display(ohlcv_pl.select(cs.numeric() | cs.temporal()).head())

Numeric OR temporal

id	date	open	high	low	close	adj_close	volume	dividends	stock_splits
i64	date	f64	f64	f64	f64	f64	i64	f64	f64
21160	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0
21161	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0
21162	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0
21163	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0
21164	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0

display(Markdown("**Invert a selector (everything NOT numeric):**"))
display(ohlcv_pl.select(~cs.numeric()).head())

Invert a selector (everything NOT numeric)

symbol	date	is_filled
str	date	bool
ABI.BR	2021-01-04	false
ABI.BR	2021-01-05	false
ABI.BR	2021-01-06	false
ABI.BR	2021-01-07	false
ABI.BR	2021-01-08	false

---

Renaming Columns

Pandas | rename()

• Rename: Rename columns.

Renames open to Open and close to Close in ohlcv_pd using a columns dictionary — the result confirms both columns are capitalised while the remaining 10 columns are unchanged.

display(
    ohlcv_pd.rename(columns={"open": "Open", "close": "Close"}).head(3)
)

	id	symbol	date	Open	high	low	Close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False

Polars | rename()

• Rename: Rename columns.

Applies the same two-column rename as the Pandas example using a plain dict (no columns= keyword) — confirms API parity while highlighting that Polars rename() takes the mapping as the first positional argument.

display(
    ohlcv_pl.rename({"open": "Open", "close": "Close"}).head(3)
)

id	symbol	date	Open	high	low	Close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

Polars | alias() inside select()

• pl.col: Reference a column by name. The foundation of all Polars expressions.
• Alias: Give an expression result a column name (Polars).

Selects symbol and date unchanged, renames close to closing_price and volume to vol using .alias() per expression — reducing the result from 12 to 4 named columns without a separate rename step.

display(
    ohlcv_pl.select(
        pl.col("symbol"),
        pl.col("date"),
        pl.col("close").alias("closing_price"),
        pl.col("volume").alias("vol"),
    ).head()
)

symbol	date	closing_price	vol
str	date	f64	i64
ABI.BR	2021-01-04	57.21	1513937
ABI.BR	2021-01-05	57.18	1382722
ABI.BR	2021-01-06	58.77	1370204
ABI.BR	2021-01-07	58.4	1469911
ABI.BR	2021-01-08	57.86	1428681

Polars | name.prefix() / name.suffix()

• Selector: Numeric: Select all numeric columns (Polars selectors module).

Adds a num_ prefix to all 9 numeric column names using cs.numeric().name.prefix(), then demonstrates pl.all().name.suffix("_raw") to append _raw to all 12 column names — showing both approaches to bulk column renaming via name modifiers.

display(Markdown("**Add prefix to numeric columns:**"))
display(
    ohlcv_pl.select(cs.numeric().name.prefix("num_")).head(3)
)

Add prefix to numeric columns

num_id	num_open	num_high	num_low	num_close	num_adj_close	num_volume	num_dividends	num_stock_splits
i64	f64	f64	f64	f64	f64	i64	f64	f64
21160	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0
21161	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0
21162	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0

display(Markdown("**Add suffix to all columns:**"))
display(
    ohlcv_pl.select(pl.all().name.suffix("_raw")).head(3)
)

Add suffix to all columns

id_raw	symbol_raw	date_raw	open_raw	high_raw	low_raw	close_raw	adj_close_raw	volume_raw	dividends_raw	stock_splits_raw	is_filled_raw
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

---

Reordering Columns

Pandas | explicit list

Reorders ohlcv_pd to a 7-column subset with date and symbol first, followed by OHLCV price columns — using a manually defined list as the column index to achieve reordering and column subsetting in one bracket operation.

new_order = ["date", "symbol", "close", "open", "high", "low", "volume"]
display(ohlcv_pd[new_order].head(3))

	date	symbol	close	open	high	low	volume
0	2021-01-04	ABI.BR	57.21	58.15	58.85	56.78	1513937
1	2021-01-05	ABI.BR	57.18	56.90	57.98	56.75	1382722
2	2021-01-06	ABI.BR	58.77	57.96	58.94	57.39	1370204

Polars | select() reorders

Reorders ohlcv_pl to the same 7-column subset as the Pandas example by passing the desired column order directly to select() — demonstrating that Polars select() naturally reorders and subsets in one step.

display(
    ohlcv_pl.select("date", "symbol", "close", "open", "high", "low", "volume").head(3)
)

date	symbol	close	open	high	low	volume
date	str	f64	f64	f64	f64	i64
2021-01-04	ABI.BR	57.21	58.15	58.85	56.78	1513937
2021-01-05	ABI.BR	57.18	56.9	57.98	56.75	1382722
2021-01-06	ABI.BR	58.77	57.96	58.94	57.39	1370204

Polars | move specific columns to front

Moves date and symbol to the front of ohlcv_pl while preserving all 12 columns by splitting the column list into front and rest, then concatenating them as the select() argument.

front = ["date", "symbol"]
rest  = [c for c in ohlcv_pl.columns if c not in front]
display(ohlcv_pl.select(front + rest).head(3))

date	symbol	id	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
date	str	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool
2021-01-04	ABI.BR	21160	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
2021-01-05	ABI.BR	21161	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
2021-01-06	ABI.BR	21162	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

---

Dropping Columns

Pandas | drop()

Removes the volume column from ohlcv_pd using drop(columns=[...]), returning an 11-column DataFrame, then drops both volume and open to produce a 10-column result.

display(ohlcv_pd.drop(columns=["volume"]).head(3))

	id	symbol	date	open	high	low	close	adj_close	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	0.0	0.0	False

display(Markdown("**Drop multiple columns:**"))
display(ohlcv_pd.drop(columns=["volume", "open"]).head(3))

Drop multiple columns

	id	symbol	date	high	low	close	adj_close	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.85	56.78	57.21	53.5761	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	57.98	56.75	57.18	53.5480	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	58.94	57.39	58.77	55.0370	0.0	0.0	False

Polars | drop()

Removes volume from ohlcv_pl using drop() with a bare string (no list required for a single column), returning an 11-column DataFrame — then drops both volume and open to confirm drop("a", "b") variadic syntax.

display(ohlcv_pl.drop("volume").head(3))

id	symbol	date	open	high	low	close	adj_close	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	0.0	0.0	false

display(Markdown("**Drop multiple columns:**"))
display(ohlcv_pl.drop("volume", "open").head(3))

Drop multiple columns

id	symbol	date	high	low	close	adj_close	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.85	56.78	57.21	53.5761	0.0	0.0	false
21161	ABI.BR	2021-01-05	57.98	56.75	57.18	53.548	0.0	0.0	false
21162	ABI.BR	2021-01-06	58.94	57.39	58.77	55.037	0.0	0.0	false

---

Selection Patterns on the Dimension Table

display(Markdown("**Dimension table columns:**"))
print("Pandas:\n", dim_pd.columns.tolist())
print("\nPolars:\n", dim_pl.columns)

Dimension table columns

Pandas: [‘id’, ‘_index’, ‘symbol’, ‘long_name’, ‘short_name’, ‘sector’, ‘sector_key’, ‘industry’, ‘industry_key’, ‘country’, ‘city’, ‘website’, ‘long_business_summary’, ‘exchange’, ‘full_exchange_name’, ‘exchange_timezone_name’, ‘exchange_timezone_short’, ‘currency’, ‘financial_currency’, ‘quote_type’, ‘market’, ‘range_start’, ‘price_data_start’, ‘valid_from’, ‘valid_to’, ‘is_current’]

Polars: [‘id’, ‘_index’, ‘symbol’, ‘long_name’, ‘short_name’, ‘sector’, ‘sector_key’, ‘industry’, ‘industry_key’, ‘country’, ‘city’, ‘website’, ‘long_business_summary’, ‘exchange’, ‘full_exchange_name’, ‘exchange_timezone_name’, ‘exchange_timezone_short’, ‘currency’, ‘financial_currency’, ‘quote_type’, ‘market’, ‘range_start’, ‘price_data_start’, ‘valid_from’, ‘valid_to’, ‘is_current’]

display(Markdown("**Pandas — select string columns:**"))
display(dim_pd.select_dtypes(include="object").drop(columns="long_business_summary").head())

Pandas | select string columns

	_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market	range_start	price_data_start	valid_to
0	euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Materials	semiconductor-equipment-materials	Netherlands	Veldhoven	https://www.asml.com	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market	1998-07-20	2021-01-01	None
1	euro_stoxx_50	MC.PA	LVMH Moët Hennessy - Louis Vuitton, Société Européenne	LVMH	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://www.lvmh.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	None
2	euro_stoxx_50	RMS.PA	Hermès International Société en commandite par actions	HERMES INTL	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://finance.hermes.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	None
3	euro_stoxx_50	OR.PA	L'Oréal S.A.	L'OREAL	Consumer Defensive	consumer-defensive	Household & Personal Products	household-personal-products	France	Clichy	https://www.loreal.com	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market	2000-01-03	2021-01-01	None
4	euro_stoxx_50	SAP.DE	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market	1998-04-09	2021-01-01	None

display(Markdown("**Polars — select string columns with cs.string():**"))
display(dim_pl.select(cs.string()).head())

Polars | select string columns with cs.string()

_index	symbol	long_name	short_name	sector	sector_key	industry	industry_key	country	city	website	long_business_summary	exchange	full_exchange_name	exchange_timezone_name	exchange_timezone_short	currency	financial_currency	quote_type	market
str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str	str
euro_stoxx_50	ASML.AS	ASML Holding N.V.	ASML HOLDING	Technology	technology	Semiconductor Equipment & Mate…	semiconductor-equipment-materi…	Netherlands	Veldhoven	https://www.asml.com	ASML Holding N.V. provides lit…	AMS	Amsterdam	Europe/Amsterdam	CET	EUR	EUR	EQUITY	nl_market
euro_stoxx_50	MC.PA	LVMH Moët Hennessy - Louis Vui…	LVMH	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://www.lvmh.com	LVMH Moët Hennessy - Louis Vui…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market
euro_stoxx_50	RMS.PA	Hermès International Société e…	HERMES INTL	Consumer Cyclical	consumer-cyclical	Luxury Goods	luxury-goods	France	Paris	https://finance.hermes.com	Hermès International Société e…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market
euro_stoxx_50	OR.PA	L'Oréal S.A.	L'OREAL	Consumer Defensive	consumer-defensive	Household & Personal Products	household-personal-products	France	Clichy	https://www.loreal.com	L'Oréal S.A., through its subs…	PAR	Paris	Europe/Paris	CET	EUR	EUR	EQUITY	fr_market
euro_stoxx_50	SAP.DE	SAP SE	SAP SE	Technology	technology	Software - Application	software-application	Germany	Walldorf	https://www.sap.com	SAP SE, together with its subs…	GER	XETRA	Europe/Berlin	CET	EUR	EUR	EQUITY	de_market

display(Markdown("**Polars — select with cs.matches() regex:**"))
display(dim_pl.select(cs.matches(".*name.*|.*id.*")).head())

Polars | select with cs.matches() regex

id	long_name	short_name	full_exchange_name	exchange_timezone_name	valid_from	valid_to
i64	str	str	str	str	datetime[ns]	null
1	ASML Holding N.V.	ASML HOLDING	Amsterdam	Europe/Amsterdam	2026-03-04 22:11:36.189862	null
2	LVMH Moët Hennessy - Louis Vui…	LVMH	Paris	Europe/Paris	2026-03-04 22:11:36.189862	null
3	Hermès International Société e…	HERMES INTL	Paris	Europe/Paris	2026-03-04 22:11:36.193940	null
4	L'Oréal S.A.	L'OREAL	Paris	Europe/Paris	2026-03-04 22:11:36.193940	null
5	SAP SE	SAP SE	XETRA	Europe/Berlin	2026-03-04 22:11:36.193940	null

---

Comparison Table | Pandas vs Polars

comparison = """
| Task | Pandas | Polars |
|---|---|---|
| Single column (Series) | `df["col"]` | `df.get_column("col")` |
| Single column (DataFrame) | `df[["col"]]` | `df.select("col")` |
| Multiple columns | `df[["a","b"]]` | `df.select("a","b")` |
| Label-based slice | `df.loc[:, "a":"c"]` | (use explicit list) |
| Position-based | `df.iloc[:, 0:3]` | `df[:, 0:3]` |
| All columns | `df` | `df.select(pl.all())` |
| Exclude columns | `df.drop(columns=[...])` | `df.select(pl.exclude(...))` |
| Numeric columns | `df.select_dtypes("number")` | `df.select(cs.numeric())` |
| String columns | `df.select_dtypes("object")` | `df.select(cs.string())` |
| Temporal columns | `df.select_dtypes("datetime")` | `df.select(cs.temporal())` |
| By specific dtype | `df.select_dtypes(include=...)` | `df.select(cs.by_dtype(...))` |
| By name pattern | `df.filter(regex=...)` | `df.select(pl.col("^regex$"))` |
| Selector by name | (manual list) | `cs.by_name("a","b")` |
| Selector set ops | (not available) | `cs.numeric() - cs.by_dtype(pl.Int64)` |
| Rename | `df.rename(columns={...})` | `df.rename({...})` |
| Alias in expr | (not applicable) | `pl.col("x").alias("y")` |
| Prefix / suffix | `df.add_prefix("p_")` | `cs.numeric().name.prefix("p_")` |
| Reorder | `df[new_order]` | `df.select(new_order)` |
| Drop columns | `df.drop(columns=[...])` | `df.drop("a","b")` |
| Regex select | `df.filter(regex="pattern")` | `pl.col("^pattern$")` |
"""
display(Markdown(comparison))

Task	Pandas	Polars
Single column (Series)	df["col"]	df.get_column("col")
Single column (DataFrame)	df[["col"]]	df.select("col")
Multiple columns	df[["a","b"]]	df.select("a","b")
Label-based slice	df.loc[:, "a":"c"]	(use explicit list)
Position-based	df.iloc[:, 0:3]	df[:, 0:3]
All columns	df	df.select(pl.all())
Exclude columns	df.drop(columns=[...])	df.select(pl.exclude(...))
Numeric columns	df.select_dtypes("number")	df.select(cs.numeric())
String columns	df.select_dtypes("object")	df.select(cs.string())
Temporal columns	df.select_dtypes("datetime")	df.select(cs.temporal())
By specific dtype	df.select_dtypes(include=...)	df.select(cs.by_dtype(...))
By name pattern	df.filter(regex=...)	df.select(pl.col("^regex$"))
Selector by name	(manual list)	cs.by_name("a","b")
Selector set ops	(not available)	cs.numeric() - cs.by_dtype(pl.Int64)
Rename	df.rename(columns={...})	df.rename({...})
Alias in expr	(not applicable)	pl.col("x").alias("y")
Prefix / suffix	df.add_prefix("p_")	cs.numeric().name.prefix("p_")
Reorder	df[new_order]	df.select(new_order)
Drop columns	df.drop(columns=[...])	df.drop("a","b")
Regex select	df.filter(regex="pattern")	pl.col("^pattern$")

---

End of notebook.

---

Filtering Rows

Row filtering selects subsets of rows based on conditions. Pandas uses boolean indexing (df[mask]), .loc[], and .query(). Polars uses .filter() with expressions. Both support compound conditions, membership tests, range checks, null filtering, and string/datetime accessors.

Performance | Vectorized expressions vs boolean masks

Polars .filter(pl.col("x") > 100) compiles into a vectorized query plan — the engine processes entire columns at once. Pandas df[df["x"] > 100] creates an intermediate boolean mask array in memory. For large datasets, Polars filtering is significantly faster and more memory-efficient.

Setup & Data Loading

print("ohlcv  :\n", ohlcv_pd.shape, "\n", list(ohlcv_pd.columns))
print("\ndim    :\n", dim_pd.shape, "\n", list(dim_pd.columns))
print("\nscores :\n", scores_pd.shape, "\n", list(scores_pd.columns))

ohlcv : (66355, 12) [‘id’, ‘symbol’, ‘date’, ‘open’, ‘high’, ‘low’, ‘close’, ‘adj_close’, ‘volume’, ‘dividends’, ‘stock_splits’, ‘is_filled’]

dim : (169, 26) [‘id’, ‘_index’, ‘symbol’, ‘long_name’, ‘short_name’, ‘sector’, ‘sector_key’, ‘industry’, ‘industry_key’, ‘country’, ‘city’, ‘website’, ‘long_business_summary’, ‘exchange’, ‘full_exchange_name’, ‘exchange_timezone_name’, ‘exchange_timezone_short’, ‘currency’, ‘financial_currency’, ‘quote_type’, ‘market’, ‘range_start’, ‘price_data_start’, ‘valid_from’, ‘valid_to’, ‘is_current’]

scores : (466, 36) [‘id’, ‘_index’, ‘symbol’, ‘score_date’, ‘sector’, ‘pe_zscore’, ‘pb_zscore’, ‘ev_ebitda_zscore’, ‘yield_zscore’, ‘relative_value_score’, ‘relative_value_rank’, ‘relative_strength’, ‘sma_50_ratio’, ‘sma_200_ratio’, ‘dist_from_52w_high’, ‘momentum_score’, ‘momentum_rank’, ‘implied_upside’, ‘recommendation_mean’, ‘price_falling_analysts_bullish’, ‘sentiment_score’, ‘sentiment_rank’, ‘composite_score’, ‘composite_rank’, ‘_scored_at’, ‘sma_30_close’, ‘sma_90_close’, ‘market_cap’, ‘index_weight’, ‘short_name’, ‘country’, ‘current_price’, ‘day_change_pct’, ‘five_day_change_pct’, ‘ytd_change_pct’, ‘currency’]

ohlcv_pd.head(3)

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False

ohlcv_pl.head(3)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

---

Boolean Indexing (Single Condition)

Chained indexing in Pandas

Chained indexing in Pandas — df[condition]["col"] = value silently fails df[df["close"] > 50]["close"] = 0 looks like it works but modifies a copy, not the original DataFrame. Pandas raises SettingWithCopyWarning but the change is lost. Always use .loc[] for assignment: df.loc[df["close"] > 50, "close"] = 0.

Polars has no chained indexing — all operations return new DataFrames, eliminating this entire class of bugs.

for all conditional assignment in Pandas

Replace any chained write (df[mask]["col"] = val) with a single .loc[] call: df.loc[df["close"] > 50, "close"] = 0. In Pandas 3+, Copy-on-Write is the default and chained assignment raises a hard error — migrating to .loc[] now is future-proof. In Polars, use pl.when(condition).then(value).otherwise(pl.col("col")) inside with_columns.

Pandas | Boolean Indexing — bracket notation

Filters ohlcv_pd to rows where the close column exceeds 50 using bracket-notation boolean indexing — the simplest Pandas filter form, returning the first 5 matching OHLCV rows.

# Rows where Close > 50
ohlcv_pd[ohlcv_pd["close"] > 50].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

Pandas | .loc with a boolean mask

Applies the same close > 50 filter as bracket notation but using .loc[], producing identical results — .loc[] is preferred over bracket notation for any conditional assignment operation to avoid SettingWithCopyWarning.

ohlcv_pd.loc[ohlcv_pd["close"] > 50].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

Polars | filter

• pl.col: Reference a column by name. The foundation of all Polars expressions.

Filters ohlcv_pl to rows where close > 50 using a pl.col expression inside .filter() — the Polars equivalent of Pandas bracket-notation boolean indexing, returning the first 5 matching rows without the SettingWithCopyWarning risk.

ohlcv_pl.filter(pl.col("close") > 50).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

---

Multiple Conditions (&, |, ~)

Pandas | AND / OR / NOT

Builds three boolean masks on ohlcv_pd: AND (close > 50 and volume > 1M), OR (close < 10 or close > 100), and NOT (volume not above 5M) — each stored in a variable and applied with .loc[] to demonstrate all three Pandas boolean operators.

# AND: close > 50 AND volume > 1_000_000
mask = (ohlcv_pd["close"] > 50) & (ohlcv_pd["volume"] > 1_000_000)
ohlcv_pd.loc[mask].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

# OR: close < 10 OR close > 100
mask = (ohlcv_pd["close"] < 10) | (ohlcv_pd["close"] > 100)
ohlcv_pd.loc[mask].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
2662	62088	ADS.DE	2021-01-04	300.0	300.5	293.0	295.4	282.2904	440364	0.0	0.0	False
2663	62089	ADS.DE	2021-01-05	292.9	295.4	288.2	289.6	276.7479	436591	0.0	0.0	False
2664	62090	ADS.DE	2021-01-06	290.7	292.7	286.8	291.7	278.7546	392602	0.0	0.0	False
2665	62091	ADS.DE	2021-01-07	294.0	294.1	288.5	288.5	275.6967	362809	0.0	0.0	False
2666	62092	ADS.DE	2021-01-08	292.3	296.8	292.0	295.1	282.0038	425762	0.0	0.0	False

# NOT: rows where volume is NOT above 5_000_000
mask = ~(ohlcv_pd["volume"] > 5_000_000)
ohlcv_pd.loc[mask].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

Polars | AND / OR / NOT

• pl.col: Reference a column by name. The foundation of all Polars expressions.

Demonstrates AND, OR, and NOT filtering in Polars using pl.col expressions inside .filter() — mirrors the Pandas examples with the same threshold values, applying &, |, and ~ operators directly in the expression context without intermediate mask variables.

ohlcv_pl.filter(
    (pl.col("close") > 50) & (pl.col("volume") > 1_000_000)
).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

ohlcv_pl.filter(
    (pl.col("close") < 10) | (pl.col("close") > 100)
).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
62088	ADS.DE	2021-01-04	300.0	300.5	293.0	295.4	282.2904	440364	0.0	0.0	false
62089	ADS.DE	2021-01-05	292.9	295.4	288.2	289.6	276.7479	436591	0.0	0.0	false
62090	ADS.DE	2021-01-06	290.7	292.7	286.8	291.7	278.7546	392602	0.0	0.0	false
62091	ADS.DE	2021-01-07	294.0	294.1	288.5	288.5	275.6967	362809	0.0	0.0	false
62092	ADS.DE	2021-01-08	292.3	296.8	292.0	295.1	282.0038	425762	0.0	0.0	false

ohlcv_pl.filter(
    ~(pl.col("volume") > 5_000_000)
).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

---

query() (Pandas Only)

ohlcv_pd.query("close > 50 and volume > 1_000_000").head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

# Using variables with @
threshold = 80
ohlcv_pd.query("close > @threshold").head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
2662	62088	ADS.DE	2021-01-04	300.0	300.5	293.0	295.4	282.2904	440364	0.0	0.0	False
2663	62089	ADS.DE	2021-01-05	292.9	295.4	288.2	289.6	276.7479	436591	0.0	0.0	False
2664	62090	ADS.DE	2021-01-06	290.7	292.7	286.8	291.7	278.7546	392602	0.0	0.0	False
2665	62091	ADS.DE	2021-01-07	294.0	294.1	288.5	288.5	275.6967	362809	0.0	0.0	False
2666	62092	ADS.DE	2021-01-08	292.3	296.8	292.0	295.1	282.0038	425762	0.0	0.0	False

# String column comparisons in query
tickers = ["SIE.DE", "SAP.DE"]
ohlcv_pd.query("symbol in @tickers").head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
55738	5301	SAP.DE	2021-01-04	108.10	108.50	104.78	105.32	97.0102	2928515	0.0	0.0	False
55739	5302	SAP.DE	2021-01-05	104.98	106.20	104.46	105.04	96.7523	2798888	0.0	0.0	False
55740	5303	SAP.DE	2021-01-06	105.14	106.26	103.60	105.48	97.1576	3018802	0.0	0.0	False
55741	5304	SAP.DE	2021-01-07	105.58	105.70	104.04	104.52	96.2734	3176143	0.0	0.0	False
55742	5305	SAP.DE	2021-01-08	105.14	106.72	105.04	106.18	97.8024	3068744	0.0	0.0	False

---

isin / is_in

Pandas | filter by ticker list with isin

Defines a list of three German ticker symbols and filters ohlcv_pd to only rows matching those tickers using .isin() — returning the first 5 OHLCV rows for SIE.DE, SAP.DE, or BAS.DE.

target_tickers = ["SIE.DE", "SAP.DE", "BAS.DE"]
ohlcv_pd[ohlcv_pd["symbol"].isin(target_tickers)].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
11965	52834	BAS.DE	2021-01-04	65.48	66.07	64.43	64.89	47.4862	2741508	0.0	0.0	False
11966	52835	BAS.DE	2021-01-05	64.30	65.47	63.26	64.40	47.1277	2770337	0.0	0.0	False
11967	52836	BAS.DE	2021-01-06	65.24	67.55	65.14	67.37	49.3011	5187251	0.0	0.0	False
11968	52837	BAS.DE	2021-01-07	67.93	68.53	67.13	68.41	50.0622	3655366	0.0	0.0	False
11969	52838	BAS.DE	2021-01-08	69.00	69.24	68.01	68.58	50.1866	3035733	0.0	0.0	False

Polars | is_in

Applies the same three-ticker filter as the Pandas example using pl.col("symbol").is_in(target_tickers) inside .filter() — reusing the same target list to confirm parity between Pandas .isin() and Polars .is_in().

target_tickers = ["SIE.DE", "SAP.DE", "BAS.DE"]
ohlcv_pl.filter(pl.col("symbol").is_in(target_tickers)).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
52834	BAS.DE	2021-01-04	65.48	66.07	64.43	64.89	47.4862	2741508	0.0	0.0	false
52835	BAS.DE	2021-01-05	64.3	65.47	63.26	64.4	47.1277	2770337	0.0	0.0	false
52836	BAS.DE	2021-01-06	65.24	67.55	65.14	67.37	49.3011	5187251	0.0	0.0	false
52837	BAS.DE	2021-01-07	67.93	68.53	67.13	68.41	50.0622	3655366	0.0	0.0	false
52838	BAS.DE	2021-01-08	69.0	69.24	68.01	68.58	50.1866	3035733	0.0	0.0	false

---

between / is_between

Pandas | close price in range 40–60 with between

Filters ohlcv_pd to rows where the close price falls in the closed interval [40, 60] using .between() — targets the lower-priced tier of the Euro Stoxx 50 universe, returning the first 5 matching rows.

ohlcv_pd[ohlcv_pd["close"].between(40, 60)].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

Polars | is_between

Demonstrates is_between() for both a numeric range (close between 40 and 60) and a date range (date between 2023-01-01 and 2023-06-30), using pl.lit(...).str.to_date() to convert string literals to dates for the second filter.

ohlcv_pl.filter(pl.col("close").is_between(40, 60)).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

# Date range filtering — Polars
ohlcv_pl.filter(
    pl.col("date").is_between(
        pl.lit("2023-01-01").str.to_date(),
        pl.lit("2023-06-30").str.to_date(),
    )
).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21675	ABI.BR	2023-01-02	56.63	57.09	56.43	56.9	54.2453	608437	0.0	0.0	false
21676	ABI.BR	2023-01-03	56.79	57.76	56.72	56.84	54.1881	1164809	0.0	0.0	false
21677	ABI.BR	2023-01-04	56.96	58.02	56.94	58.02	55.313	1835512	0.0	0.0	false
21678	ABI.BR	2023-01-05	57.69	57.98	57.0	57.14	54.4741	1324250	0.0	0.0	false
21679	ABI.BR	2023-01-06	57.23	57.47	57.03	57.44	54.7601	1100010	0.0	0.0	false

---

Null / NaN Filtering

Pandas | Null / NaN Filtering

Filters scores_pd to rows where ev_ebitda_zscore is null using .isna(), then to rows where it is not null using .notna() — identifies which index constituents lack EV/EBITDA data (financial institutions without an enterprise value ratio).

# Rows where a column IS null
scores_pd[scores_pd["ev_ebitda_zscore"].isna()].head()

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
0	163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.261140	NaN	2.388962	1.521163	1	0.016123	1.009090	1.130264	0.082486	0.477966	16	0.153157	1.84211	False	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085000	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.320	0.011437	-0.073156	0.105582	EUR
8	188	euro_stoxx_50	SAN.MC	2026-03-04	Financial Services	0.458599	0.499398	NaN	-1.107590	-0.049864	33	0.393197	0.953824	1.139519	0.113499	0.519307	14	0.224704	1.70000	False	0.448776	15	0.306073	9	2026-03-04 22:40:25.489180	10.604900	9.919500	145955749888	0.028570	BANCO SANTANDER S.A.	Spain	9.982	0.038818	-0.105876	-0.008739	EUR
15	176	euro_stoxx_50	ISP.MI	2026-03-04	Financial Services	0.366619	0.488302	NaN	0.723361	0.526094	13	-0.065247	0.922492	0.990855	0.119662	-0.122626	33	0.254150	2.00000	False	0.146959	21	0.183476	16	2026-03-04 22:40:25.489180	5.834933	5.769367	94268317696	0.018452	INTESA SANPAOLO	Italy	5.422	0.018216	-0.067103	-0.084276	EUR
16	195	euro_stoxx_50	UCG.MI	2026-03-04	Financial Services	0.452501	0.355225	NaN	-0.260674	0.182351	26	0.085867	0.950889	1.054430	0.137862	0.123670	22	0.261521	1.94444	False	0.240819	18	0.182280	17	2026-03-04 22:40:25.489180	73.253333	68.983556	103066501120	0.020174	UNICREDIT	Italy	68.790	0.027483	-0.072161	-0.030034	EUR
20	175	euro_stoxx_50	INGA.AS	2026-03-04	Financial Services	0.379917	0.727879	NaN	-0.240509	0.289096	21	0.113086	0.946227	1.070815	0.118737	0.192271	21	0.201459	2.10526	False	-0.145505	30	0.111954	21	2026-03-04 22:40:25.489180	24.766000	23.632333	67454382080	0.013204	ING GROEP N.V.	Netherlands	23.305	0.018798	-0.066867	-0.029363	EUR

# Rows where a column is NOT null
scores_pd[scores_pd["ev_ebitda_zscore"].notna()].head()

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
1	168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.120650	1.112416	0.055524	0.685752	8	0.121212	1.33333	False	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838000	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.000	0.011649	-0.019608	0.193059	EUR
2	174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.37500	False	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.066490	0.164723	EUR
3	172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.051850	2.541889	1	0.075269	1.80000	False	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675000	129.122000	139207262208	0.027249	Siemens Energy AG	Germany	162.750	0.047297	-0.039256	0.351744	EUR
4	149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	False	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342000	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.480	-0.017073	-0.040762	0.174499	EUR
5	196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.379830	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	NaN	False	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.600	0.013786	-0.048756	-0.090390	EUR

Polars | Null / NaN Filtering

• pl.col: Reference a column by name. The foundation of all Polars expressions.

Filters scores_pl to null rows with .is_null() and non-null rows with .is_not_null() on ev_ebitda_zscore — the Polars equivalents of Pandas .isna() / .notna(), confirming the same financial institutions are identified as lacking EV/EBITDA data.

scores_pl.filter(pl.col("ev_ebitda_zscore").is_null()).head()

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.26114	null	2.388962	1.521163	1	0.016123	1.00909	1.130264	0.082486	0.477966	16	0.153157	1.84211	false	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
188	euro_stoxx_50	SAN.MC	2026-03-04	Financial Services	0.458599	0.499398	null	-1.10759	-0.049864	33	0.393197	0.953824	1.139519	0.113499	0.519307	14	0.224704	1.7	false	0.448776	15	0.306073	9	2026-03-04 22:40:25.489180	10.6049	9.9195	145955749888	0.02857	BANCO SANTANDER S.A.	Spain	9.982	0.038818	-0.105876	-0.008739	EUR
176	euro_stoxx_50	ISP.MI	2026-03-04	Financial Services	0.366619	0.488302	null	0.723361	0.526094	13	-0.065247	0.922492	0.990855	0.119662	-0.122626	33	0.25415	2.0	false	0.146959	21	0.183476	16	2026-03-04 22:40:25.489180	5.834933	5.769367	94268317696	0.018452	INTESA SANPAOLO	Italy	5.422	0.018216	-0.067103	-0.084276	EUR
195	euro_stoxx_50	UCG.MI	2026-03-04	Financial Services	0.452501	0.355225	null	-0.260674	0.182351	26	0.085867	0.950889	1.05443	0.137862	0.12367	22	0.261521	1.94444	false	0.240819	18	0.18228	17	2026-03-04 22:40:25.489180	73.253333	68.983556	103066501120	0.020174	UNICREDIT	Italy	68.79	0.027483	-0.072161	-0.030034	EUR
175	euro_stoxx_50	INGA.AS	2026-03-04	Financial Services	0.379917	0.727879	null	-0.240509	0.289096	21	0.113086	0.946227	1.070815	0.118737	0.192271	21	0.201459	2.10526	false	-0.145505	30	0.111954	21	2026-03-04 22:40:25.489180	24.766	23.632333	67454382080	0.013204	ING GROEP N.V.	Netherlands	23.305	0.018798	-0.066867	-0.029363	EUR

scores_pl.filter(pl.col("ev_ebitda_zscore").is_not_null()).head()

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.12065	1.112416	0.055524	0.685752	8	0.121212	1.33333	false	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.0	0.011649	-0.019608	0.193059	EUR
174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.375	false	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.06649	0.164723	EUR
172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.05185	2.541889	1	0.075269	1.8	false	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675	129.122	139207262208	0.027249	Siemens Energy AG	Germany	162.75	0.047297	-0.039256	0.351744	EUR
149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	false	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.48	-0.017073	-0.040762	0.174499	EUR
196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	null	false	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.6	0.013786	-0.048756	-0.09039	EUR

---

where / mask (Pandas) vs when / then / otherwise (Polars)

These do not strictly filter rows — they replace values conditionally while keeping all rows.

Pandas | where — keep values where True, replace with NaN where False

Applies .where() to the close column, keeping the original price where volume > 1M and replacing it with NaN otherwise — demonstrating conditional value preservation that retains all rows while masking low-volume prices.

# Keep close where volume > 1M, else NaN
ohlcv_pd["close"].where(ohlcv_pd["volume"] > 1_000_000).head(10)

	close
0	57.21
1	57.18
2	58.77
3	58.40
4	57.86
5	56.61
6	56.51
7	56.48
8	56.96
9	56.74

Pandas | mask — opposite of where (replace where True)

Applies .mask() to the close column, replacing prices with NaN where volume > 5M — the inverse of .where(), masking out high-volume spikes rather than preserving low-volume entries.

# Replace close with NaN where volume > 5M
ohlcv_pd["close"].mask(ohlcv_pd["volume"] > 5_000_000).head(10)

	close
0	57.21
1	57.18
2	58.77
3	58.40
4	57.86
5	56.61
6	56.51
7	56.48
8	56.96
9	56.74

Polars | when / then / otherwise

• With Columns: Add new columns or replace existing ones. All original columns are kept.
• pl.col: Reference a column by name. The foundation of all Polars expressions.
• pl.lit: Create a constant/literal value as an expression.

Creates a close_filtered column keeping close where volume > 1M and None otherwise, then chains multiple when/then to bucket prices into “high”, “mid”, or “low” — both computed as new columns via with_columns() without dropping any rows.

ohlcv_pl.with_columns(
    pl.when(pl.col("volume") > 1_000_000)
      .then(pl.col("close"))
      .otherwise(pl.lit(None))
      .alias("close_filtered")
).select("symbol", "date", "close", "volume", "close_filtered").head(10)

symbol	date	close	volume	close_filtered
str	date	f64	i64	f64
ABI.BR	2021-01-04	57.21	1513937	57.21
ABI.BR	2021-01-05	57.18	1382722	57.18
ABI.BR	2021-01-06	58.77	1370204	58.77
ABI.BR	2021-01-07	58.4	1469911	58.4
ABI.BR	2021-01-08	57.86	1428681	57.86
ABI.BR	2021-01-11	56.61	1518079	56.61
ABI.BR	2021-01-12	56.51	1649991	56.51
ABI.BR	2021-01-13	56.48	1090806	56.48
ABI.BR	2021-01-14	56.96	1523045	56.96
ABI.BR	2021-01-15	56.74	1769988	56.74

# Multiple conditions with when/then chaining (Polars)
ohlcv_pl.with_columns(
    pl.when(pl.col("close") > 100)
      .then(pl.lit("high"))
      .when(pl.col("close") > 50)
      .then(pl.lit("mid"))
      .otherwise(pl.lit("low"))
      .alias("price_bucket")
).select("symbol", "date", "close", "price_bucket").head(10)

symbol	date	close	price_bucket
str	date	f64	str
ABI.BR	2021-01-04	57.21	mid
ABI.BR	2021-01-05	57.18	mid
ABI.BR	2021-01-06	58.77	mid
ABI.BR	2021-01-07	58.4	mid
ABI.BR	2021-01-08	57.86	mid
ABI.BR	2021-01-11	56.61	mid
ABI.BR	2021-01-12	56.51	mid
ABI.BR	2021-01-13	56.48	mid
ABI.BR	2021-01-14	56.96	mid
ABI.BR	2021-01-15	56.74	mid

---

String Accessor Filtering

Pandas | String Accessor Filtering — .str

• String Ops: Text manipulation via .str accessor: contains, split, replace, extract.

Filters ohlcv_pd via the .str accessor: first to tickers starting with “S” using .str.startswith() (e.g., SAF.PA, SAP.DE, SAN.MC), then to tickers containing “DE” using .str.contains() — selecting German-exchange stocks by exchange suffix.

# Tickers that start with "S"
ohlcv_pd[ohlcv_pd["symbol"].str.startswith("S")].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
51747	18512	SAF.PA	2021-01-04	117.35	121.00	116.1	116.15	111.6447	658764	0.0	0.0	False
51748	18513	SAF.PA	2021-01-05	114.90	116.95	114.8	116.40	111.8850	588765	0.0	0.0	False
51749	18514	SAF.PA	2021-01-06	117.55	117.55	115.4	116.30	111.7888	581543	0.0	0.0	False
51750	18515	SAF.PA	2021-01-07	117.05	117.30	114.8	115.80	111.3082	635605	0.0	0.0	False
51751	18516	SAF.PA	2021-01-08	116.90	117.00	115.0	116.35	111.8369	688460	0.0	0.0	False

# Tickers containing "DE"
ohlcv_pd[ohlcv_pd["symbol"].str.contains("DE")].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
2662	62088	ADS.DE	2021-01-04	300.0	300.5	293.0	295.4	282.2904	440364	0.0	0.0	False
2663	62089	ADS.DE	2021-01-05	292.9	295.4	288.2	289.6	276.7479	436591	0.0	0.0	False
2664	62090	ADS.DE	2021-01-06	290.7	292.7	286.8	291.7	278.7546	392602	0.0	0.0	False
2665	62091	ADS.DE	2021-01-07	294.0	294.1	288.5	288.5	275.6967	362809	0.0	0.0	False
2666	62092	ADS.DE	2021-01-08	292.3	296.8	292.0	295.1	282.0038	425762	0.0	0.0	False

Polars | String Accessor Filtering — .str

• String Ops: Text manipulation via .str accessor: contains, split, replace, extract.
• pl.col: Reference a column by name. The foundation of all Polars expressions.

Applies the same string filters as the Pandas example using str.starts_with() and str.contains() inside pl.col(...).filter() — note starts_with (no underscore) vs Pandas’ startswith, returning identical ticker subsets.

ohlcv_pl.filter(pl.col("symbol").str.starts_with("S")).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
18512	SAF.PA	2021-01-04	117.35	121.0	116.1	116.15	111.6447	658764	0.0	0.0	false
18513	SAF.PA	2021-01-05	114.9	116.95	114.8	116.4	111.885	588765	0.0	0.0	false
18514	SAF.PA	2021-01-06	117.55	117.55	115.4	116.3	111.7888	581543	0.0	0.0	false
18515	SAF.PA	2021-01-07	117.05	117.3	114.8	115.8	111.3082	635605	0.0	0.0	false
18516	SAF.PA	2021-01-08	116.9	117.0	115.0	116.35	111.8369	688460	0.0	0.0	false

ohlcv_pl.filter(pl.col("symbol").str.contains("DE")).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
62088	ADS.DE	2021-01-04	300.0	300.5	293.0	295.4	282.2904	440364	0.0	0.0	false
62089	ADS.DE	2021-01-05	292.9	295.4	288.2	289.6	276.7479	436591	0.0	0.0	false
62090	ADS.DE	2021-01-06	290.7	292.7	286.8	291.7	278.7546	392602	0.0	0.0	false
62091	ADS.DE	2021-01-07	294.0	294.1	288.5	288.5	275.6967	362809	0.0	0.0	false
62092	ADS.DE	2021-01-08	292.3	296.8	292.0	295.1	282.0038	425762	0.0	0.0	false

---

Datetime Accessor Filtering

Pandas | Datetime Accessor Filtering — .dt

• DateTime Accessor: Extract date parts: .dt.year(), .dt.month(), .dt.weekday().
• Parse Dates: Convert strings to datetime objects (Pandas).

Converts the date column to datetime with pd.to_datetime() (required in Pandas before .dt access), then filters to January rows using .dt.month == 1 — demonstrating date-part extraction for seasonal or calendar-based filtering.

# Ensure date is datetime
ohlcv_pd["date"] = pd.to_datetime(ohlcv_pd["date"])
 
# Filter rows in January
ohlcv_pd[ohlcv_pd["date"].dt.month == 1].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

# Filter by year
ohlcv_pd[ohlcv_pd["date"].dt.year == 2023].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
515	21675	ABI.BR	2023-01-02	56.63	57.09	56.43	56.90	54.2453	608437	0.0	0.0	False
516	21676	ABI.BR	2023-01-03	56.79	57.76	56.72	56.84	54.1881	1164809	0.0	0.0	False
517	21677	ABI.BR	2023-01-04	56.96	58.02	56.94	58.02	55.3130	1835512	0.0	0.0	False
518	21678	ABI.BR	2023-01-05	57.69	57.98	57.00	57.14	54.4741	1324250	0.0	0.0	False
519	21679	ABI.BR	2023-01-06	57.23	57.47	57.03	57.44	54.7601	1100010	0.0	0.0	False

# Filter by day of week (Monday=0)
ohlcv_pd[ohlcv_pd["date"].dt.dayofweek == 0].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
5	21165	ABI.BR	2021-01-11	57.73	57.81	56.39	56.61	53.0142	1518079	0.0	0.0	False
10	21170	ABI.BR	2021-01-18	56.25	57.30	56.20	57.08	53.4544	730298	0.0	0.0	False
15	21175	ABI.BR	2021-01-25	54.77	54.80	52.89	53.17	49.7927	1974547	0.0	0.0	False
20	21180	ABI.BR	2021-02-01	52.32	53.30	52.15	52.58	49.2402	1543605	0.0	0.0	False

Polars | Datetime Accessor Filtering — .dt

• DateTime Accessor: Extract date parts: .dt.year(), .dt.month(), .dt.weekday().
• pl.col: Reference a column by name. The foundation of all Polars expressions.

Filters ohlcv_pl by month (January), year (2023), and weekday (Monday = 1 in Polars) using pl.col("date").dt.method() — note Polars requires function-call syntax (.dt.month()) where Pandas uses attribute access (.dt.month).

ohlcv_pl.filter(pl.col("date").dt.month() == 1).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

ohlcv_pl.filter(pl.col("date").dt.year() == 2023).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21675	ABI.BR	2023-01-02	56.63	57.09	56.43	56.9	54.2453	608437	0.0	0.0	false
21676	ABI.BR	2023-01-03	56.79	57.76	56.72	56.84	54.1881	1164809	0.0	0.0	false
21677	ABI.BR	2023-01-04	56.96	58.02	56.94	58.02	55.313	1835512	0.0	0.0	false
21678	ABI.BR	2023-01-05	57.69	57.98	57.0	57.14	54.4741	1324250	0.0	0.0	false
21679	ABI.BR	2023-01-06	57.23	57.47	57.03	57.44	54.7601	1100010	0.0	0.0	false

ohlcv_pl.filter(pl.col("date").dt.weekday() == 1).head()  # Monday=1 in Polars

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21165	ABI.BR	2021-01-11	57.73	57.81	56.39	56.61	53.0142	1518079	0.0	0.0	false
21170	ABI.BR	2021-01-18	56.25	57.3	56.2	57.08	53.4544	730298	0.0	0.0	false
21175	ABI.BR	2021-01-25	54.77	54.8	52.89	53.17	49.7927	1974547	0.0	0.0	false
21180	ABI.BR	2021-02-01	52.32	53.3	52.15	52.58	49.2402	1543605	0.0	0.0	false

---

head, tail, slice, sample

# Pandas
print("head(3):\n", ohlcv_pd.head(3), "\n")
print("tail(3):\n", ohlcv_pd.tail(3))

head(3): id symbol date open high low close adj_close volume
0 21160 ABI.BR 2021-01-04 58.15 58.85 56.78 57.21 53.5761 1513937 1 21161 ABI.BR 2021-01-05 56.90 57.98 56.75 57.18 53.5480 1382722 2 21162 ABI.BR 2021-01-06 57.96 58.94 57.39 58.77 55.0370 1370204

dividends stock_splits is_filled 0 0.0 0.0 False 1 0.0 0.0 False 2 0.0 0.0 False

tail(3): id symbol date open high low close adj_close
66352 66876 WKL.AS 2026-03-10 68.8 69.16 66.34 67.16 67.16 66353 66877 WKL.AS 2026-03-11 67.5 69.60 67.02 67.22 67.22 66354 66929 WKL.AS 2026-03-12 67.0 67.54 66.28 67.32 67.32

volume dividends stock_splits is_filled 66352 1355645 0.0 0.0 False 66353 1142531 0.0 0.0 False 66354 210379 0.0 0.0 False

# Pandas — iloc slicing
ohlcv_pd.iloc[10:15]

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
10	21170	ABI.BR	2021-01-18	56.25	57.30	56.20	57.08	53.4544	730298	0.0	0.0	False
11	21171	ABI.BR	2021-01-19	57.10	57.26	56.26	56.35	52.7707	1116570	0.0	0.0	False
12	21172	ABI.BR	2021-01-20	56.35	56.77	56.00	56.24	52.6677	1226516	0.0	0.0	False
13	21173	ABI.BR	2021-01-21	56.20	56.55	55.31	55.31	51.7968	1404283	0.0	0.0	False
14	21174	ABI.BR	2021-01-22	55.28	55.28	54.12	54.78	51.3005	1557287	0.0	0.0	False

# Pandas — sample
ohlcv_pd.sample(5, random_state=42)

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
43053	38920	MUV2.DE	2023-03-29	320.0000	322.600	318.400	322.4000	290.3200	195809	0.0	0.0	False
56209	5772	SAP.DE	2022-11-03	95.9200	96.340	95.080	95.5100	91.9052	1424973	0.0	0.0	False
53515	11015	SAN.MC	2022-09-15	2.5975	2.686	2.597	2.6765	2.3373	70158349	0.0	0.0	False
6498	28954	AI.PA	2025-08-12	173.3200	174.440	172.800	173.6800	173.6800	415652	0.0	0.0	False
63527	24956	UCG.MI	2025-07-07	56.4600	57.350	56.440	57.3500	56.0468	4705567	0.0	0.0	False

# Polars
ohlcv_pl.head(3)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false

ohlcv_pl.tail(3)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
66876	WKL.AS	2026-03-10	68.8	69.16	66.34	67.16	67.16	1355645	0.0	0.0	false
66877	WKL.AS	2026-03-11	67.5	69.6	67.02	67.22	67.22	1142531	0.0	0.0	false
66929	WKL.AS	2026-03-12	67.0	67.54	66.28	67.32	67.32	210379	0.0	0.0	false

ohlcv_pl.slice(10, 5)  # offset, length

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21170	ABI.BR	2021-01-18	56.25	57.3	56.2	57.08	53.4544	730298	0.0	0.0	false
21171	ABI.BR	2021-01-19	57.1	57.26	56.26	56.35	52.7707	1116570	0.0	0.0	false
21172	ABI.BR	2021-01-20	56.35	56.77	56.0	56.24	52.6677	1226516	0.0	0.0	false
21173	ABI.BR	2021-01-21	56.2	56.55	55.31	55.31	51.7968	1404283	0.0	0.0	false
21174	ABI.BR	2021-01-22	55.28	55.28	54.12	54.78	51.3005	1557287	0.0	0.0	false

ohlcv_pl.sample(5, seed=42)

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
11529	SAN.MC	2024-09-18	4.511	4.5455	4.5065	4.5085	4.2785	16487238	0.0	0.0	false
35604	CS.PA	2025-10-14	39.34	40.27	39.25	40.18	40.18	3511125	0.0	0.0	false
63666	WKL.AS	2022-01-05	102.2	102.65	101.25	101.8	95.254	230509	0.0	0.0	false
33136	PRX.AS	2021-05-03	41.3654	41.737	41.0718	41.3746	40.8581	2177525	0.0	0.0	false
19417	SAF.PA	2024-07-12	204.2	204.8	201.3	204.8	202.5157	496739	0.0	0.0	false

---

unique / drop_duplicates / drop_nulls / dropna

# Pandas — unique tickers
ohlcv_pd["symbol"].drop_duplicates().head(10)

	symbol
0	ABI.BR
1331	AD.AS
2662	ADS.DE
3986	ADYEN.AS
5317	AI.PA
6648	AIR.PA
7979	ALV.DE
9303	ARGX.BR
10634	ASML.AS
11965	BAS.DE

# Pandas — drop_duplicates on subset
ohlcv_pd.drop_duplicates(subset=["symbol"]).head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.1500	58.8500	56.7800	57.2100	53.5761	1513937	0.0	0.0	False
1331	59438	AD.AS	2021-01-04	23.3800	23.8300	23.3800	23.7900	19.9339	3526165	0.0	0.0	False
2662	62088	ADS.DE	2021-01-04	300.0000	300.5000	293.0000	295.4000	282.2904	440364	0.0	0.0	False
3986	60763	ADYEN.AS	2021-01-04	1900.0000	1921.5000	1856.0000	1859.5000	1859.5000	99408	0.0	0.0	False
5317	27773	AI.PA	2021-01-04	112.3554	113.6364	111.9835	112.7686	102.9625	917982	0.0	0.0	False

# Pandas — dropna
scores_pd.dropna(subset=["ev_ebitda_zscore"]).head()

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
1	168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.120650	1.112416	0.055524	0.685752	8	0.121212	1.33333	False	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838000	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.000	0.011649	-0.019608	0.193059	EUR
2	174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.37500	False	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.066490	0.164723	EUR
3	172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.051850	2.541889	1	0.075269	1.80000	False	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675000	129.122000	139207262208	0.027249	Siemens Energy AG	Germany	162.750	0.047297	-0.039256	0.351744	EUR
4	149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	False	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342000	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.480	-0.017073	-0.040762	0.174499	EUR
5	196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.379830	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	NaN	False	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.600	0.013786	-0.048756	-0.090390	EUR

# Polars — unique
ohlcv_pl.select("symbol").unique().head(10)

symbol
str
ARGX.BR
OR.PA
IBE.MC
ADYEN.AS
ENI.MI
BAYN.DE
IFX.DE
BNP.PA
DSY.PA
ENEL.MI

# Polars — unique on subset
ohlcv_pl.unique(subset=["symbol"]).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
17194	ENR.DE	2021-01-04	30.62	31.2	29.91	30.13	29.8481	1693265	0.0	0.0	false
1	ASML.AS	2021-01-04	404.0	411.0	402.25	406.25	387.709	789502	0.0	0.0	false
63406	WKL.AS	2021-01-04	69.78	71.3	69.76	70.86	65.1785	516176	0.0	0.0	false
42307	ENI.MI	2021-01-04	8.604	8.756	8.397	8.448	6.0466	19734004	0.0	0.0	false
19837	IBE.MC	2021-01-04	11.8	11.945	11.79	11.905	9.4733	14213672	0.0	0.0	false

# Polars — drop_nulls
scores_pl.drop_nulls(subset=["ev_ebitda_zscore"]).head()

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.12065	1.112416	0.055524	0.685752	8	0.121212	1.33333	false	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.0	0.011649	-0.019608	0.193059	EUR
174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.375	false	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.06649	0.164723	EUR
172	euro_stoxx_50	ENR.DE	2026-03-04	Industrials	-0.902738	-0.743338	-1.693212	-1.326075	-1.166341	46	1.645455	1.137007	1.474095	0.05185	2.541889	1	0.075269	1.8	false	-0.123264	29	0.417428	4	2026-03-04 22:40:25.489180	155.675	129.122	139207262208	0.027249	Siemens Energy AG	Germany	162.75	0.047297	-0.039256	0.351744	EUR
149	euro_stoxx_50	ABI.BR	2026-03-04	Consumer Defensive	0.474084	0.844075	0.552739	-0.975005	0.223973	25	-0.029791	1.058542	1.142783	0.063063	0.651891	10	0.186198	1.69231	false	0.344755	17	0.406873	5	2026-03-04 22:40:25.489180	64.342	57.869333	125566156800	0.024579	AB INBEV	Belgium	64.48	-0.017073	-0.040762	0.174499	EUR
196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	null	false	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.6	0.013786	-0.048756	-0.09039	EUR

---

Sorting

Pandas | sort_values

Sorts ohlcv_pd by close descending to surface the highest-priced OHLCV rows (Hermès RMS.PA at ~2839), then by ["symbol", "date"] with mixed ascending/descending order to list the most recent date first within each ticker.

ohlcv_pd.sort_values("close", ascending=False).head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
51473	3708	RMS.PA	2025-02-14	2926.0	2957.0	2813.0	2839.0	2802.9382	105651	0.0	0.0	False
51472	3707	RMS.PA	2025-02-13	2770.0	2816.0	2765.0	2816.0	2780.2302	80087	0.0	0.0	False
51474	3709	RMS.PA	2025-02-17	2825.0	2858.0	2803.0	2809.0	2776.7424	53852	3.5	0.0	False
51475	3710	RMS.PA	2025-02-18	2816.0	2827.0	2780.0	2806.0	2773.7771	65469	0.0	0.0	False
4150	60927	ADYEN.AS	2021-08-24	2725.0	2766.0	2711.5	2766.0	2766.0000	61431	0.0	0.0	False

# Multi-column sort
ohlcv_pd.sort_values(["symbol", "date"], ascending=[True, False]).head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
1330	66897	ABI.BR	2026-03-12	62.64	62.96	62.08	62.76	62.76	303648	0.0	0.0	False
1329	66781	ABI.BR	2026-03-11	62.64	63.38	62.38	62.68	62.68	1679807	0.0	0.0	False
1328	66780	ABI.BR	2026-03-10	62.74	63.34	62.24	63.32	63.32	1828703	0.0	0.0	False
1327	66779	ABI.BR	2026-03-09	61.72	62.70	61.50	62.58	62.58	1880254	0.0	0.0	False
1326	64764	ABI.BR	2026-03-06	63.46	63.64	62.32	63.14	63.14	2519869	0.0	0.0	False

Polars | sort

Sorts ohlcv_pl by close descending, then by ["symbol", "date"] with mixed directions, and demonstrates sort_by inside a with_columns / over expression to reorder close values chronologically within each symbol partition — a pattern useful in group-aware calculations.

ohlcv_pl.sort("close", descending=True).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
3708	RMS.PA	2025-02-14	2926.0	2957.0	2813.0	2839.0	2802.9382	105651	0.0	0.0	false
3707	RMS.PA	2025-02-13	2770.0	2816.0	2765.0	2816.0	2780.2302	80087	0.0	0.0	false
3709	RMS.PA	2025-02-17	2825.0	2858.0	2803.0	2809.0	2776.7424	53852	3.5	0.0	false
3710	RMS.PA	2025-02-18	2816.0	2827.0	2780.0	2806.0	2773.7771	65469	0.0	0.0	false
60927	ADYEN.AS	2021-08-24	2725.0	2766.0	2711.5	2766.0	2766.0	61431	0.0	0.0	false

# Multi-column sort
ohlcv_pl.sort(["symbol", "date"], descending=[False, True]).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
66897	ABI.BR	2026-03-12	62.64	62.96	62.08	62.76	62.76	303648	0.0	0.0	false
66781	ABI.BR	2026-03-11	62.64	63.38	62.38	62.68	62.68	1679807	0.0	0.0	false
66780	ABI.BR	2026-03-10	62.74	63.34	62.24	63.32	63.32	1828703	0.0	0.0	false
66779	ABI.BR	2026-03-09	61.72	62.7	61.5	62.58	62.58	1880254	0.0	0.0	false
64764	ABI.BR	2026-03-06	63.46	63.64	62.32	63.14	63.14	2519869	0.0	0.0	false

# sort_by inside an expression context (useful in group_by)
ohlcv_pl.with_columns(
    pl.col("close").sort_by("date").over("symbol").alias("close_chronological")
).head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled	close_chronological
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false	57.21
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false	57.18
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false	58.77
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false	58.4
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false	57.86

---

Filtering After a Join (Practical Example)

Filter OHLCV data to only include tickers that appear in the dimension table.

# Pandas — semi-join style filter
valid_tickers = dim_pd["symbol"].unique()
ohlcv_pd[ohlcv_pd["symbol"].isin(valid_tickers)].head()

	id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
0	21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	False
1	21161	ABI.BR	2021-01-05	56.90	57.98	56.75	57.18	53.5480	1382722	0.0	0.0	False
2	21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.0370	1370204	0.0	0.0	False
3	21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.40	54.6905	1469911	0.0	0.0	False
4	21164	ABI.BR	2021-01-08	58.16	58.40	57.43	57.86	54.1848	1428681	0.0	0.0	False

# Polars — semi join
ohlcv_pl.join(dim_pl.select("symbol").unique(), on="symbol", how="semi").head()

id	symbol	date	open	high	low	close	adj_close	volume	dividends	stock_splits	is_filled
i64	str	date	f64	f64	f64	f64	f64	i64	f64	f64	bool
21160	ABI.BR	2021-01-04	58.15	58.85	56.78	57.21	53.5761	1513937	0.0	0.0	false
21161	ABI.BR	2021-01-05	56.9	57.98	56.75	57.18	53.548	1382722	0.0	0.0	false
21162	ABI.BR	2021-01-06	57.96	58.94	57.39	58.77	55.037	1370204	0.0	0.0	false
21163	ABI.BR	2021-01-07	58.68	58.86	57.88	58.4	54.6905	1469911	0.0	0.0	false
21164	ABI.BR	2021-01-08	58.16	58.4	57.43	57.86	54.1848	1428681	0.0	0.0	false

---

Filtering scores_daily — Practical Examples

scores_pd.head(3)

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
0	163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.261140	NaN	2.388962	1.521163	1	0.016123	1.009090	1.130264	0.082486	0.477966	16	0.153157	1.84211	False	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085000	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.320	0.011437	-0.073156	0.105582	EUR
1	168	euro_stoxx_50	DTE.DE	2026-03-04	Communication Services	0.326587	0.387463	0.379532	-0.127867	0.241429	24	-0.205598	1.120650	1.112416	0.055524	0.685752	8	0.121212	1.33333	False	0.617835	10	0.515005	2	2026-03-04 22:40:25.489180	30.838000	28.554556	164294311936	0.032159	DEUTSCHE TELEKOM AG	Germany	33.000	0.011649	-0.019608	0.193059	EUR
2	174	euro_stoxx_50	IFX.DE	2026-03-04	Technology	0.509398	0.637215	0.677068	-0.696662	0.281755	22	0.000965	1.048244	1.198626	0.088845	0.675764	9	0.126408	1.37500	False	0.579187	11	0.512235	3	2026-03-04 22:40:25.489180	43.480333	38.855556	57222533120	0.011201	INFINEON TECHNOLOGIES AG	Germany	43.945	0.054343	-0.066490	0.164723	EUR

# Pandas — top scores
scores_pd[scores_pd["relative_value_score"] > 0.8].head()

	id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
0	163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.261140	NaN	2.388962	1.521163	1	0.016123	1.009090	1.130264	0.082486	0.477966	16	0.153157	1.84211	False	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085000	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
5	196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.379830	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	NaN	False	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.60	0.013786	-0.048756	-0.090390	EUR
7	166	euro_stoxx_50	DG.PA	2026-03-04	Industrials	0.778573	0.850985	0.928797	1.146268	0.926156	3	-0.031697	1.064353	1.095809	0.062871	0.595820	11	0.043608	2.04762	False	-0.538059	34	0.327972	8	2026-03-04 22:40:25.489180	131.013333	123.067778	74446422016	0.014572	VINCI	France	134.15	0.006754	-0.054283	0.117451	EUR
18	191	euro_stoxx_50	SGO.PA	2026-03-04	Industrials	1.060288	0.991397	1.009773	0.546912	0.902092	4	-0.390177	0.899452	0.848740	0.276512	-0.946419	43	0.360809	1.94737	True	0.530945	13	0.162206	19	2026-03-04 22:40:25.489180	86.202000	85.090222	38256795648	0.007488	SAINT GOBAIN	France	77.16	-0.011783	-0.121185	-0.112695	EUR
19	189	euro_stoxx_50	SAN.PA	2026-03-04	Healthcare	0.799367	0.646615	0.714066	1.090175	0.812556	7	-0.432698	0.982603	0.949547	0.285444	-0.592341	39	0.262148	2.00000	True	0.170637	20	0.130284	20	2026-03-04 22:40:25.489180	79.889667	82.906444	95673352192	0.018727	SANOFI	France	79.23	-0.007889	-0.019309	-0.042191	EUR

# Polars — top scores
scores_pl.filter(pl.col("relative_value_score") > 0.8).head()

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.26114	null	2.388962	1.521163	1	0.016123	1.00909	1.130264	0.082486	0.477966	16	0.153157	1.84211	false	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	null	false	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.6	0.013786	-0.048756	-0.09039	EUR
166	euro_stoxx_50	DG.PA	2026-03-04	Industrials	0.778573	0.850985	0.928797	1.146268	0.926156	3	-0.031697	1.064353	1.095809	0.062871	0.59582	11	0.043608	2.04762	false	-0.538059	34	0.327972	8	2026-03-04 22:40:25.489180	131.013333	123.067778	74446422016	0.014572	VINCI	France	134.15	0.006754	-0.054283	0.117451	EUR
191	euro_stoxx_50	SGO.PA	2026-03-04	Industrials	1.060288	0.991397	1.009773	0.546912	0.902092	4	-0.390177	0.899452	0.84874	0.276512	-0.946419	43	0.360809	1.94737	true	0.530945	13	0.162206	19	2026-03-04 22:40:25.489180	86.202	85.090222	38256795648	0.007488	SAINT GOBAIN	France	77.16	-0.011783	-0.121185	-0.112695	EUR
189	euro_stoxx_50	SAN.PA	2026-03-04	Healthcare	0.799367	0.646615	0.714066	1.090175	0.812556	7	-0.432698	0.982603	0.949547	0.285444	-0.592341	39	0.262148	2.0	true	0.170637	20	0.130284	20	2026-03-04 22:40:25.489180	79.889667	82.906444	95673352192	0.018727	SANOFI	France	79.23	-0.007889	-0.019309	-0.042191	EUR

# Polars — chain multiple filters
scores_pl.filter(
    pl.col("relative_value_score").is_not_null(),
    pl.col("relative_value_score") > 0.5,
).head()

id	_index	symbol	score_date	sector	pe_zscore	pb_zscore	ev_ebitda_zscore	yield_zscore	relative_value_score	relative_value_rank	relative_strength	sma_50_ratio	sma_200_ratio	dist_from_52w_high	momentum_score	momentum_rank	implied_upside	recommendation_mean	price_falling_analysts_bullish	sentiment_score	sentiment_rank	composite_score	composite_rank	_scored_at	sma_30_close	sma_90_close	market_cap	index_weight	short_name	country	current_price	day_change_pct	five_day_change_pct	ytd_change_pct	currency
i64	str	str	date	str	f64	f64	f64	f64	f64	i64	f64	f64	f64	f64	f64	i64	f64	f64	bool	f64	i64	f64	i64	datetime[ns]	f64	f64	i64	f64	str	str	f64	f64	f64	f64	str
163	euro_stoxx_50	BNP.PA	2026-03-04	Financial Services	0.913389	1.26114	null	2.388962	1.521163	1	0.016123	1.00909	1.130264	0.082486	0.477966	16	0.153157	1.84211	false	0.052711	25	0.683947	1	2026-03-04 22:40:25.489180	92.085	81.181889	99751215104	0.019525	BNP PARIBAS ACT.A	France	89.32	0.011437	-0.073156	0.105582	EUR
196	euro_stoxx_50	VOW.DE	2026-03-04	Consumer Cyclical	1.166221	0.940273	0.37983	1.528891	1.003804	2	-0.292748	0.929392	0.969628	0.180805	-0.411969	37	0.297071	null	false	0.555357	12	0.382397	6	2026-03-04 22:40:25.489180	102.286667	101.225556	47923826688	0.009381	VOLKSWAGEN AG	Germany	95.6	0.013786	-0.048756	-0.09039	EUR
194	euro_stoxx_50	TTE.PA	2026-03-04	Energy	0.691106	0.609377	0.44961	0.731948	0.62051	12	0.0498	1.109161	1.212503	0.08411	0.919444	5	0.041131	2.04545	false	-0.542576	35	0.332459	7	2026-03-04 22:40:25.489180	63.603	58.231667	142003961856	0.027796	TOTALENERGIES	France	66.86	-0.018209	-0.00757	0.202734	EUR
166	euro_stoxx_50	DG.PA	2026-03-04	Industrials	0.778573	0.850985	0.928797	1.146268	0.926156	3	-0.031697	1.064353	1.095809	0.062871	0.59582	11	0.043608	2.04762	false	-0.538059	34	0.327972	8	2026-03-04 22:40:25.489180	131.013333	123.067778	74446422016	0.014572	VINCI	France	134.15	0.006754	-0.054283	0.117451	EUR
150	euro_stoxx_50	AD.AS	2026-03-04	Consumer Defensive	0.758806	0.330473	0.741401	1.0574	0.72202	9	0.035485	1.155305	1.170693	0.008379	1.135547	4	-0.014969	2.29412	false	-1.03108	44	0.275495	12	2026-03-04 22:40:25.489180	37.249	35.750667	36734455808	0.00719	KONINKLIJKE AHOLD DELHAIZE N.V…	Netherlands	41.42	0.019946	0.007786	0.187841	EUR

---

Comparison Table | Filtering Rows

comparison = r"""
| Operation                        | Pandas                                         | Polars                                          |
|:---------------------------------|:-------------------------------------------------|:-------------------------------------------------|
| Single boolean filter            | `df[df["col"] > x]`                             | `df.filter(pl.col("col") > x)`                   |
| `.loc` with condition            | `df.loc[mask]`                                   | `df.filter(mask_expr)`                           |
| AND / OR / NOT                   | `(c1) & (c2)`, `(c1) \| (c2)`, `~c`              | Same operators on expressions                    |
| `query()`                        | `df.query("col > 5")`                            | N/A — use `filter` expressions                   |
| `isin` / `is_in`                | `df[df["col"].isin(lst)]`                        | `df.filter(pl.col("col").is_in(lst))`             |
| `between` / `is_between`        | `df[df["col"].between(a, b)]`                    | `df.filter(pl.col("col").is_between(a, b))`       |
| Null check                       | `df[df["col"].isna()]` / `.notna()`               | `filter(pl.col("col").is_null())` / `.is_not_null()` |
| `where` / `mask`                | `s.where(cond)` / `s.mask(cond)`                  | `when(cond).then(val).otherwise(alt)`            |
| String filter (`startswith`)     | `df[df["c"].str.startswith("X")]`                 | `filter(pl.col("c").str.starts_with("X"))`        |
| Datetime filter (month)         | `df[df["d"].dt.month == 1]`                       | `filter(pl.col("d").dt.month() == 1)`             |
| `head` / `tail`                 | `df.head(n)` / `df.tail(n)`                       | Same                                             |
| Slice                            | `df.iloc[a:b]`                                    | `df.slice(offset, length)`                       |
| Sample                           | `df.sample(n, random_state=42)`                   | `df.sample(n, seed=42)`                          |
| Unique rows                      | `df.drop_duplicates(subset=…)`                    | `df.unique(subset=…)`                            |
| Drop nulls                       | `df.dropna(subset=…)`                             | `df.drop_nulls(subset=…)`                        |
| Sort                             | `df.sort_values("col", ascending=False)`           | `df.sort("col", descending=True)`                 |
| Multi-col sort                   | `sort_values(["a","b"], ascending=[T,F])`          | `sort(["a","b"], descending=[F,T])`               |
| Expression-level sort            | N/A                                               | `pl.col("c").sort_by("d").over("g")`              |
| Semi-join filter                 | `df[df["k"].isin(other["k"])]`                    | `df.join(other, on="k", how="semi")`              |
"""
display(Markdown(comparison))

Operation	Pandas	Polars
Single boolean filter	df[df["col"] > x]	df.filter(pl.col("col") > x)
.loc with condition	df.loc[mask]	df.filter(mask_expr)
AND / OR / NOT	(c1) & (c2), (c1) | (c2), ~c	Same operators on expressions
query()	df.query("col > 5")	N/A — use filter expressions
isin / is_in	df[df["col"].isin(lst)]	df.filter(pl.col("col").is_in(lst))
between / is_between	df[df["col"].between(a, b)]	df.filter(pl.col("col").is_between(a, b))
Null check	df[df["col"].isna()] / .notna()	filter(pl.col("col").is_null()) / .is_not_null()
where / mask	s.where(cond) / s.mask(cond)	when(cond).then(val).otherwise(alt)
String filter (startswith)	df[df["c"].str.startswith("X")]	filter(pl.col("c").str.starts_with("X"))
Datetime filter (month)	df[df["d"].dt.month == 1]	filter(pl.col("d").dt.month() == 1)
head / tail	df.head(n) / df.tail(n)	Same
Slice	df.iloc[a:b]	df.slice(offset, length)
Sample	df.sample(n, random_state=42)	df.sample(n, seed=42)
Unique rows	df.drop_duplicates(subset=…)	df.unique(subset=…)
Drop nulls	df.dropna(subset=…)	df.drop_nulls(subset=…)
Sort	df.sort_values("col", ascending=False)	df.sort("col", descending=True)
Multi-col sort	sort_values(["a","b"], ascending=[T,F])	sort(["a","b"], descending=[F,T])
Expression-level sort	N/A	pl.col("c").sort_by("d").over("g")
Semi-join filter	df[df["k"].isin(other["k"])]	df.join(other, on="k", how="semi")

---

Common Traps and Safe Patterns

Pandas filter() vs Polars filter()

The same method name means different things across libraries

In Pandas, df.filter(items=[...]) selects columns by label. In Polars, df.filter(expr) selects rows by expression. Reusing the name without checking the API leads to quiet, wrong results.

Use the row-selection API that matches the library

In Pandas, filter rows with boolean masks or .loc[...]. In Polars, use .filter(...) with expressions such as pl.col("close") > 50.

Parenthesize Boolean Conditions

Unparenthesized boolean expressions change operator precedence

df[df["a"] > 5 & df["b"] < 10] is parsed incorrectly because & binds more tightly than the comparison operators. The same precedence trap exists in both Pandas and Polars expression code.

Wrap each condition and combine them with bitwise operators

Write (cond1) & (cond2) and (cond1) | (cond2) explicitly. That makes the intent unambiguous and keeps the filter semantics correct in both libraries.

Nulls Need Null Predicates

Equality filters never match missing values

In Pandas, NaN != NaN, so df["col"] == value cannot recover null rows. The same conceptual rule applies elsewhere: missing values need dedicated null predicates, not equality comparisons.

Use .isna() or .is_null() when the target is missing data

Reach for df["col"].isna() in Pandas and pl.col("col").is_null() in Polars whenever the filtering condition is “missing” rather than “equal to a concrete value”.

Label Slices and Positional Slices Behave Differently

.loc and .iloc do not share the same endpoint rules

.loc["a":"c"] includes both labels, while .iloc[0:3] excludes the right endpoint. Swapping one for the other without adjusting the slice leads to off-by-one errors.

Decide first whether the slice is label-based or position-based

Use .loc when the boundary values are labels you want included, and .iloc when you mean Python-style positional slicing. Treat them as different tools, not as interchangeable spellings.

Deduplication Depends on Existing Order

keep=\"first\" is only meaningful after deterministic sorting

drop_duplicates(..., keep="first") and similar “first row wins” patterns are order-sensitive. If the incoming frame is unsorted, the surviving row is arbitrary from a business perspective.

Sort first when deduplication must be reproducible

Establish the winning row explicitly with a sort on timestamp, priority, or another business key before you drop duplicates or keep the first occurrence.

Sampling Without a Seed Breaks Reproducibility

Unseeded samples change on every run

Random sampling without random_state= in Pandas or seed= in Polars returns different rows every time. That makes notebooks, tests, and benchmarks harder to compare or debug.

Set the sampling seed whenever the result must be repeatable

Pass random_state= in Pandas and seed= in Polars for any sample that will be inspected, committed, tested, or compared across runs.

Python Explore, Select and Filter Recommendations

1. Profile before transforming — run describe(), null_count(), value_counts(), and n_unique() on every dataset before writing any transformation logic. This takes seconds and prevents hours of debugging.
2. Filter early, select early — push filters and column selection as close to the data source as possible. In Polars lazy mode, this enables predicate pushdown and projection pushdown.
3. Use Polars selectors for type-based selection — cs.numeric(), cs.string(), cs.temporal() are safer than hard-coding column names, which break when schemas change.
4. Use validate= on joins before filtering — if your filter depends on a join result, validate the join cardinality first (validate="one_to_one" or "many_to_one") to catch unexpected row multiplication.
5. Prefer is_in() over chained | conditions — df.filter(pl.col("ticker").is_in(["A", "B", "C"])) is cleaner and faster than (col == "A") | (col == "B") | (col == "C").
6. Always pass dropna=False in Pandas value_counts() — the default drops NaN, hiding the null count from your cardinality analysis.
7. Sort before head()/tail() in unsorted data — head() on an unsorted DataFrame returns arbitrary rows, not the “first” in any meaningful order.

Troubleshooting and failure modes

Symptom	Likely cause	Fix
KeyError on column selection	Column name has whitespace or case mismatch	df.columns = df.columns.str.strip() then verify exact names
Filter returns empty DataFrame unexpectedly	Filter condition is too restrictive, or NaN rows excluded silently	Check value_counts(dropna=False) on the filter column
TypeError: Cannot perform 'rand_' with...	Missing parentheses around boolean conditions	Wrap each condition in (): (cond1) & (cond2)
.loc returns a Series instead of a DataFrame	Selecting a single column with .loc[:, "col"]	Use .loc[:, ["col"]] (list) to get a DataFrame
ColumnNotFoundError in Polars	Column name does not exist or was renamed upstream	Check df.columns or df.schema before the failing operation
describe() shows unexpected count < total rows	Non-numeric columns excluded by default (Pandas)	Use describe(include="all") or target specific dtypes
sample() gives different results across runs	No seed set	Pass random_state=42 (Pandas) or seed=42 (Polars)
unique() returns unordered values	Both libraries return unique values in arbitrary order	Chain .sort() after unique() if order matters
Polars filter() returns all rows unchanged	Expression always evaluates to True (e.g., comparing wrong column)	Print the boolean expression separately to verify: df.select(expr)
isin() returns all False	List values don’t match column dtype (e.g., string “1” vs integer 1)	Ensure the list values match the column dtype exactly