SplitCheck.get_tabular_report()

Get a tabular report of the checks of the splits.

Usage

Source

SplitCheck.get_tabular_report(title=None)

The report has a group of rows for each check: the overlap of each pair of splits, the labels of each split, and the drift of each column in each split after the first. Each row has a badge with its result, and the header says how many issues there are.

Parameters

title: str | None = None

An optional title replacing the default.

Returns

GT

A Great Tables object.

Examples

import pointblank as pb
import polars as pl

train = pl.DataFrame({"x": list(range(100)), "y": ["a", "b"] * 50})
test = pl.DataFrame({"x": list(range(50, 100)), "y": ["a"] * 50})

pb.check_splits(train=train, test=test, label="y").get_tabular_report()

Split check
train: 100 rows · test: 50 rows 4 issues found.
Check	Result
Overlap
Rows of test also in train compared on x	LEAK 50 of 50 rows (100.0%) 50 the same
Labels
Labels of train in y	BASE a 50% · b 50% 2 labels
Labels of test in y, compared with train	ISSUE a 100% missing 'b'; PSI 4.6
Drift
x in test compared with train	MAJOR PSI 3.8 numeric; KS 0.5 (p = 0)
PSI is the population stability index of the shares of the values (binned by the deciles of 'train', for a numeric column): a change is moderate from 0.1 and major from 0.2. KS is the Kolmogorov-Smirnov statistic.

Each row of the test split is also in the training split, the test split has none of the label "b", and its values of x are larger.