features: null

transpose: true

selector_grid_search: true

selector_grid_search_params:

  cv:

    splitter: StratifiedShuffleSplit

    n_splits: 5

    test_size: 0.1

  iid: false

  scoring: roc_auc

preprocess_steps:

  # apply log transformation

  - log_transform:

      name: LogTransform

      type: transformer

      enabled: true

      params:

        base: 2

        pseudo_count: 1

  # method to scale features across samples

  - scale_features:

      name: StandardScaler

      type: scaler

      enabled: true

      params:

        with_mean: true

# template for grid_search_params in classifiers

classifier_grid_search_params:

  cv:

    splitter: StratifiedShuffleSplit

    n_splits: 5

    test_size: 0.1

  iid: false

  scoring: roc_auc

classifiers:

  LogRegL2:

    classifier: LogisticRegression

    # parameters for the classifier used for feature selection

    classifier_params:

      penalty: l2

      solver: liblinear

    # grid search for hyper-parameters for the classifier

    grid_search: true

    grid_search_params:

      param_grid:

        C: [0.00001, 0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000]

  RandomForest:

    classifier: RandomForestClassifier

    grid_search: true

    grid_search_params:

      param_grid:

        n_estimators: [25, 50, 75]

        max_depth: [3, 4, 5]

  RBFSVM:

    classifier: SVC

    classifier_params:

      kernel: rbf

      gamma: scale

    grid_search: true

    grid_search_params:

      param_grid:

        C: [0.00001, 0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000]

  DecisionTree:

    classifier: DecisionTreeClassifier

    grid_search: true

    grid_search_params:

      param_grid:

        max_depth: [2, 3, 4, 5, 6, 7, 8]

  MLP:

    classifier: MLPClassifier

    classifier_params:

      activation: relu

      solver: adam

      max_iter: 40

    grid_search: true

    grid_search_params:

      param_grid:

        hidden_layer_sizes: [[50], [100], [150], [200], [250], [300]]

# cross-validation parameters for performance evaluation

cv_params:

  splitter: StratifiedShuffleSplit

  # number of train-test splits for cross-validation

  n_splits: 50

  # number or proportion of samples to use as test set

  test_size: 0.1

  # scoring metric for performance evaluation

  scoring: roc_auc

# method for computing sample weight

#  balanced: compute sample weight from data such that classes are balanced

sample_weight: balanced