Summary#
- class skore.Summary(dataframe, project=None)[source]#
Metadata and metrics for all reports persisted in a project at a given moment.
A summary object stores the metadata and metrics of the reports that have been persisted in a project as a
pandas.DataFrame, accessible through theframe()method. It implements a custom HTML representation that displays the information as a table. The table provides a per-row selection and builds a query string that can be copied and passed toquery()to recover a subset of reports.Use
query()to filter the reports andcompare()to load the corresponding report objects from the project.See also
compare()Compare selected reports side by side.
Project.summarize()Create a summary from a project.
- compare(*, return_as='list')[source]#
Return the reports referenced by the summary object from the project.
- Parameters:
- return_as{“list”, “report”}, default=”list”
In what form the reports should be returned.
- Returns:
- reportslist of reports or ComparisonReport
If
return_as="list", a list ofEstimatorReportorCrossValidationReportobjects. Ifreturn_as="report", aComparisonReport.
See also
compare()Compare reports side by side.
- frame(*, report_type=None)[source]#
Return the metadata and metrics as a
pandas.DataFrame.- Parameters:
- report_type{“estimator”, “cross-validation”}, default=None
Filter the rows to a specific type of report. When
None, all reports are returned.
- Returns:
- framepandas.DataFrame
The metadata and metrics of the reports.
- plot()[source]#
Plot a visual summary of the reports referenced by this object.
- Raises:
- NotImplementedError
This method is not implemented yet.
- query(expr)[source]#
Filter the summary using a
pandas.DataFrame.query()expression.- Parameters:
- exprstr
The query string to evaluate. The report
idis available as an index level, so selections built from the HTML representation (e.g."id in ['ab12cd34']") can be used to recover specific reports.
- Returns:
- summarySummary
A new summary restricted to the rows matching
expr.