analyze#

EstimatorReport.data.analyze(*, data_source='both', with_y=True, subsample=None, subsample_strategy='head', seed=None)[source]#

Plot dataset statistics.

Parameters:

data_source{“train”, “test”, “both”}, default=”both”

The dataset to analyze. If “train”, only the training set is used. If “test”, only the test set is used. If “both”, both sets are concatenated vertically.

with_ybool, default=True

Whether to include the target variable in the analysis. If True, the target variable is concatenated horizontally to the features.

subsampleint, default=None

The number of points to subsample the dataframe hold by the display, using the strategy set by subsample_strategy. It must be a strictly positive integer. If None, no subsampling is applied.

subsample_strategy{“head”, “random”}, default=”head”

The strategy used to subsample the dataframe hold by the display. It only has an effect when subsample is not None.

If "head": subsample by taking the subsample first points of the dataframe, similar to Pandas: df.head(n).
If "random": randomly subsample the dataframe by using a uniform distribution. The random seed is controlled by seed.

seedint, default=None

The random seed to use when randomly subsampling. It only has an effect when subsample is not None and subsample_strategy='random'.

Returns:

TableReportDisplay: A display object containing the dataset statistics and plots.

Examples

>>> from sklearn.datasets import load_breast_cancer
>>> from sklearn.linear_model import LogisticRegression
>>> from skore import evaluate
>>> X, y = load_breast_cancer(return_X_y=True)
>>> classifier = LogisticRegression(max_iter=10_000)
>>> report = evaluate(classifier, X, y, splitter=0.2, pos_label=1)
>>> report.data.analyze().frame()