Parameter settings for dataset

(create a new yaml file in ~/search/properties/dataset/your_new_dataset.yaml)

The benchmark provides several arguments for describing:

• Basic setting of the parameters

See below for the details:

Dataset

• name (str) : The name of the dataset.

• path (str) : The path to the dataset.

Column definition

• query_col (str) : The query column name; if not present, it will be added.

• id_col (str) : The ID column name.

• score_col (str) : The score column name.

• feature_cols (list) : The list of column names for the features.

• sensitive_cols (list) : The list of sensitive columns to be considered.

Train-Test split

• k_fold (int) : The number of k-fold stratified splits.

• ratio_split (float) : The ratio of the test set.

• pos_th (int) : The threshold for considering negative samples during training.