config.yaml

data_path: "./data/census.csv"
test_size: 0.2
random_seed: 42
target: "salary"
features:
  numerical:
    - "age"
    - "fnlgt"
    - "education_num"
    - "capital_gain"
    - "capital_loss"
    - "hours_per_week"
  categorical:
    - "workclass"
    - "education"
    - "marital_status"
    - "occupation"
    - "relationship"
    - "race"
    - "sex"
    - "native_country"
random_forest_parameters:
  # This section is passed to the random forest model
  # as a dictionary of parameters, thus names must match with
  # the sklearn API.
  # Whenever we have model or other object with many parameters
  # we should write config files for them.
  # That is easier than passing parameters in the code or via CLI
  # and we can guarantee compatibility in the code in case the model API changes
  # (i.e., we would simply change the config file).
  # NOTE: These default parameters can be overwritten
  # by the hyperparameter tuning 
  n_estimators: 100
  criterion: 'gini'
  max_depth: 13
  min_samples_split: 2
  min_samples_leaf: 1
  min_weight_fraction_leaf: 0.0
  max_features: 'auto'
  max_leaf_nodes: null
  min_impurity_decrease: 0.0
  bootstrap: true
  oob_score: false
  n_jobs: null
  random_state: 42
  verbose: 0
  warm_start: false
  class_weight: "balanced"
  ccp_alpha: 0.0
  max_samples: null
random_forest_grid_search:
  hyperparameters:
    n_estimators:
      - 100
      - 150
      - 200
    max_features:
      - "sqrt"
      - "log2"
    criterion:
      - "gini"
      - "entropy"
    max_depth:
      - 5
      - 10
      - 15
  cv: 3
  scoring: 'f1'
slicing_min_data_points: 5
model_artifact: "./exported_artifacts/model.pickle"
processing_artifact: "./exported_artifacts/processing_parameters.pickle"
evaluation_artifact: "./exported_artifacts/evaluation_report.txt"
slicing_artifact: "./exported_artifacts/slice_output.txt"