dvc.yaml

stages:
  get-data:
    cmd: python src/data/data_recovering.py training
    deps: 
    - src/data/data_recovering.py
    - docs/data_schema.yaml
    outs:
    - data/raw
  clean-data:
    cmd: python src/data/data_cleaning.py
    deps:
    - data/raw
    - src/data/data_cleaning.py
    outs:
    - data/interim
    params:
    - params/data-params.yaml:
  generate-features:
    cmd: python src/features/data_to_features.py
    deps:
    - data/interim
    - src/features/data_to_features.py
    outs:
    - data/features
    - docs/feature_schema.json:
        cache: false
    params:
    - params/data-params.yaml:
  train-model:
    cmd: python src/model/train_model.py
    deps:
    - data/features
    - src/model/train_model.py
    - docs/feature_schema.json
    outs:
    - trained_model
    params:
    - params/params.yaml:
        - metrics
        - loss
    - params/data-params.yaml:
        - label_column
    - params/train-params.yaml
  get-validation-data:
    cmd: python src/data/data_recovering.py validation
    deps: 
    - src/data/data_recovering.py
    outs:
    - data/validation
  validate:
    cmd: python serving/validate.py
    deps:
    - serving/validate.py
    - data/validation
    - trained_model
    params:
    - params/data-params.yaml:
    metrics:
    - metrics/validation_results.json:
        cache: false