syndi.benchmark.benchmark

syndi.benchmark.benchmark(tasks, metrics=None, agnostic_metrics=False, output_path='results/', summary_metric='accuracy', is_regression=False, preprocess_fn=None)[source]

benchmark function

Is the main function of this library that runs benchmark testing on a set of tasks. Return detailed results of each run stored in a DataFrame object.

Parameters:
  • tasks (list) – a list of task instances storing meta information of each task.

  • metrics (list) – pycaret classification metrics to record a subset of [‘Accuracy’, ‘AUC’, ‘Recall’, ‘F1’, ‘Precision’, ‘Kappa’]

  • agnostic_metrics (boolean) – whether to record dataset agnostic metrics in results

  • output_path (str) – the dir path to store benchmark results and records of each task.

  • save_results (boolean) – whether to store the benchmark results.

Returns:

benchmarking results in detail.

Return type:

pd.DataFrame