Skip to content

pricing_report

Return the rating tables, the actual versus expected cells and the diagnostics of the pruning and graduation stages, in one document for review.

The report describes the data passed to it; it is not evidence that the data were held out. Save it beside the serialized model. Regression and binary classification only.

Note

The model prediction results will be correct only if the X parameter with feature values contains all the features used in the model. For a polars DataFrame or LazyFrame, the features are matched by name: extra columns are ignored and the order of the columns does not matter (a LazyFrame collects only the columns the model needs). For other types, and for a model trained without feature names, the features must be in the same number and order as the columns provided during the training.

Method call format

pricing_report(X, y, *, sample_weight=None, exposure=None, ref_measure=None)

Parameters

X

Description

Feature values data.

Possible types

  • polars.DataFrame
  • polars.LazyFrame
  • numpy.ndarray of shape (object_count, feature_count)
  • other array-like data of the same shape

Default value

Required parameter

y

Description

The target values of the objects. A string names a column of a polars X.

Possible types

  • numpy.ndarray of shape (object_count,)
  • polars.Series
  • list
  • string

Default value

Required parameter

sample_weight, exposure

Description

The weight and the exposure of each object, passed explicitly as for actual_vs_expected.

Possible types

  • numpy.ndarray of shape (object_count,)
  • polars.Series
  • list
  • string

Default value

None

ref_measure

Description

The reference measure of the exported tables, as for tables.

Possible types

string

Default value

None

Return value

A dictionary:

  • report_version — The version of the report format.
  • prediction_units, reference_normalization, band_interpretation — How to read the numbers.
  • tables — The rating tables, as returned by tables.
  • actual_vs_expected — The cells returned by actual_vs_expected, each feature with its axis.
  • pruning — The pruning_report_ attribute.
  • graduation — The diagnostics of graduation.

Usage examples

import json

# model: a TBoostRegressor(objective="poisson") trained with exposure="Exposure"
report = model.pricing_report(train_data, "ClaimCount", exposure="Exposure")
with open("freq_report.json", "w") as f:
    json.dump(report, f)