OCDocker.OCScore.Analysis.StudyProcessing package

Parse and structure Optuna study results into best-RMSE, best-AUC, and best-combined views with consensus metrics.

Usage:

from OCDocker.OCScore.Analysis.StudyProcessing import get_study_data

OCDocker.OCScore.Analysis.StudyProcessing.get_study_data(snames, storage, final_metrics, n_trials, error_threshold=1.5, nn_ae_start=None, nn_ae_end=None, xgb_ga_start=None, xgb_ga_end=None)[source]

Retrieve Optuna study data and structure it by best RMSE, AUC, and combined metrics.

Parameters:

  • snames (list[str]) – List of study names.

  • storage (str) – SQLAlchemy storage string.

  • final_metrics (pd.DataFrame) – Consensus and raw metric dataframe.

  • n_trials (int) – Number of trials per study.

  • error_threshold (float) – Threshold to filter maximum RMSE.

  • nn_ae_start (int | None, optional) – Start index for “NN + AE” labeling.

  • nn_ae_end (int | None, optional) – End index for “NN + AE” labeling.

  • xgb_ga_start (int | None, optional) – Start index for “XGB + GA” labeling.

  • xgb_ga_end (int | None, optional) – End index for “XGB + GA” labeling.

Returns:

Filtered RMSE, AUC, combined metric dataframes + full results_df + ranges.

Return type:

tuple