definitions:
  datasets:
    implanted_ostmeyer:
      format: ImmuneML
      params:
        metadata_file: simulated_data/repertoire_implanting_rate__2e-05/sim_instruction/exported_dataset/immuneml/rep_size_metadata.csv
        path: simulated_data/repertoire_implanting_rate__2e-05/sim_instruction/exported_dataset/immuneml/rep_size.iml_dataset
  encodings:
    binarized_feature_size_2:
      KmerFrequency:
        k: 2
        normalization_type: BINARY
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    binarized_feature_size_3:
      KmerFrequency:
        k: 3
        normalization_type: BINARY
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    binarized_feature_size_4:
      KmerFrequency:
        k: 4
        normalization_type: BINARY
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    feature_size_2:
      KmerFrequency:
        k: 2
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    feature_size_3:
      KmerFrequency:
        k: 3
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    feature_size_4:
      KmerFrequency:
        k: 4
        scale_to_unit_variance: true
        scale_to_zero_mean: true
    one_hot_vanilla:
      OneHot:
        flatten: true
        sequence_type: amino_acid
        use_positional_info: false
  ml_methods:
    logistic_regression:
      LogisticRegression:
        C:
        - 1
        - 0.1
        - 0.05
        penalty: l1
      model_selection_cv: true
      model_selection_n_folds: 5
    rf:
      RandomForestClassifier:
        n_estimators:
        - 5
        - 50
        - 200
        - 2000
      model_selection_cv: true
      model_selection_n_folds: 5
  reports:
    my_hp_benchmark: MLSettingsPerformance
instructions:
  hpoptim_instr:
    assessment:
      split_count: 5
      split_strategy: k_fold
      training_percentage: 0.7
    dataset: implanted_ostmeyer
    labels:
    - cancer
    metrics:
    - auc
    - recall
    number_of_processes: 16
    optimization_metric: accuracy
    refit_optimal_model: false
    reports:
    - my_hp_benchmark
    selection:
      split_count: 3
      split_strategy: random
      training_percentage: 0.7
    settings:
    - encoding: feature_size_4
      ml_method: logistic_regression
    - encoding: feature_size_3
      ml_method: logistic_regression
    - encoding: feature_size_2
      ml_method: logistic_regression
    - encoding: binarized_feature_size_4
      ml_method: logistic_regression
    - encoding: binarized_feature_size_3
      ml_method: logistic_regression
    - encoding: binarized_feature_size_2
      ml_method: logistic_regression
    - encoding: one_hot_vanilla
      ml_method: logistic_regression
    - encoding: feature_size_4
      ml_method: rf
    - encoding: feature_size_3
      ml_method: rf
    - encoding: feature_size_2
      ml_method: rf
    - encoding: binarized_feature_size_4
      ml_method: rf
    - encoding: binarized_feature_size_3
      ml_method: rf
    - encoding: binarized_feature_size_2
      ml_method: rf
    - encoding: one_hot_vanilla
      ml_method: rf
    strategy: GridSearch
    type: TrainMLModel