load_cv_results

load_cv_results#

mlproject.postprocess.utils.load_cv_results(models_dir, model_type, target_name, feat_set_type, n_folds, collect_sizes=False)[source]#

Load cross-validation results and aggregate test MAE errors.

Parameters:

  • models_dir (str) – Base directory containing model results.

  • model_type (str) – Model name/prefix (e.g., ‘rf’, ‘modnet’).

  • target_name (str) – Target property name.

  • feat_set_type (str) – Subfolder suffix (e.g., ‘matminer’, ‘matminer_lob’).

  • n_folds (int) – Number of CV folds.

  • collect_sizes (bool, optional) – If True, also return train/test set sizes per fold.

Returns:

  • mean_test_errors (list of float) – Mean test error for each fold.

  • fold_test_errors (list of np.ndarray) – Raw test errors for each fold.

  • n_train_list (list of int (optional)) – Number of training samples per fold.

  • n_test_list (list of int (optional)) – Number of test samples per fold.

Contents