Comparing multiple reports

The class ComparisonReport provides a report allowing to compare EstimatorReport or CrossValidationReport instances in an interactive way. The functionalities of the report are accessible through accessors.

ComparisonReport(reports, *[, n_jobs])

Compare several estimator or cross-validation reports.

Methods

ComparisonReport.help()	Display report help using rich or HTML.
ComparisonReport.create_estimator_report(*, ...)	Create an estimator report from one of the reports in the comparison.
ComparisonReport.get_predictions(*, data_source)	Get predictions from the underlying reports.

Accessors

ComparisonReport.checks	Accessor for checks-related operations.
ComparisonReport.inspection	Accessor for model inspection related operations.
ComparisonReport.metrics	Accessor for metrics-related operations.

Metrics

The metrics accessor helps you to evaluate the statistical performance of the compared estimators. In addition, we provide a sub-accessor plot, to get the common performance metric representations.

Registered metrics whose name is a valid Python identifier are also available as report.metrics.<name>(...). Use available() (or help()) at runtime to discover which metrics apply to a given report.

ComparisonReport.metrics.help()	Display accessor help using rich or HTML.
ComparisonReport.metrics.summarize(*[, ...])	Report a set of metrics for the estimators.
ComparisonReport.metrics.add(metric, *[, ...])	Add a custom metric to summarize().
ComparisonReport.metrics.available(*[, ...])	List available metric names in the registry.
ComparisonReport.metrics.confusion_matrix(*)	Plot the confusion matrix.
ComparisonReport.metrics.get(name[, ...])	Get a metric value.
ComparisonReport.metrics.precision_recall(*)	Plot the precision-recall curve.
ComparisonReport.metrics.prediction_error(*)	Plot the prediction error of a regression model.
ComparisonReport.metrics.remove(name)	Remove a metric from each underlying estimator report.
ComparisonReport.metrics.roc(*[, data_source])	Plot the ROC curve.
ComparisonReport.metrics.score(*[, ...])	Compute the estimator's default score.
ComparisonReport.metrics.timings(*[, aggregate])	Get all measured processing times related to the different estimators.

Inspection

The inspection accessor helps you inspect your model by e.g. evaluating the importance of the features in your model.

ComparisonReport.inspection.help()	Display accessor help using rich or HTML.
ComparisonReport.inspection.coefficients()	Retrieve the coefficients for each report, including the intercepts.
ComparisonReport.inspection.impurity_decrease()	Retrieve the Mean Decrease in Impurity (MDI) for each report.
ComparisonReport.inspection.permutation_importance(*)	Display the permutation feature importance.

Checks

The checks accessor runs automated checks that look for common modeling problems such as overfitting and underfitting.

ComparisonReport.checks.help()	Display accessor help using rich or HTML.
ComparisonReport.checks.summarize(*[, ...])	Run checks and return a summary with detected issues.
ComparisonReport.checks.add(checks)	Register additional checks for this report.
ComparisonReport.checks.available()	List available checks in the registry.
ComparisonReport.checks.remove(code)	Remove a check from the registry.