Report Generator

July 5, 2024 ยท View on GitHub

Scikit-learn_bench report generator creates a high-level report with aggregated stats from provided benchmark results.

Generator will eventually support different types of reports, but there is only one supported type currently:

  • Separate tables: writes aggregated metrics on summary page and detailed result on separate pages

Raw results are converted into a pandas dataframe and the final report is made by processing this dataframe into separate ones written into Excel tables.

Arguments

NameTypeDefault valueChoicesDescription
--report-log-levelstrWARNING('ERROR', 'WARNING', 'INFO', 'DEBUG')Logging level for report generator.
--result-filesstr['result.json']Result file path[s] from scikit-learn_bench runs for report generation.
--report-filestrreport.xlsxReport file path.
--report-typestrseparate-tables('separate-tables',)Report type ("separate-tables" is the only supported now).
--compatibility-modeFalse[EXPERIMENTAL] Compatibility mode drops and modifies results to make them comparable (for example, sklearn and cuML parameters).
--drop-columns
--drop-cols
str[]Columns to drop from report.
--diff-columns
--diff-cols
str['environment_name', 'library', 'format', 'device']Columns to show difference between.
--split-columnsstr['estimator', 'method', 'function']Splitting columns for subreports/sheets.
--diffs-selectionstrupper_triangle['upper_triangle', 'lower_triangle', 'matrix']Selects which part of one-vs-one difference to show (all matrix or one of triangles).
--perf-color-scalefloat[0.8, 1.0, 10.0]Color scale for performance metric improvement in report.
--quality-color-scalefloat[0.99, 0.995, 1.01]Color scale for quality metric improvement in report.

Documentation tree