dvc.yaml

stages:
  preprocess:
    cmd: python src/data/preprocessing.py
    deps:
      - src/data/preprocessing.py
      - params.yaml
      - data/raw/Dev_data_to_be_shared.csv
      - data/raw/validation_data_to_be_shared.csv
    outs:
      - data/interim/train_data_cleaned.csv
      - data/interim/validation_data_cleaned.csv

  feature_engineering:
    cmd: python src/features/feature_engineering.py
    deps:
      - src/features/feature_engineering.py
      - params.yaml
      - data/interim/train_data_cleaned.csv
      - data/interim/validation_data_cleaned.csv
    outs:
      - data/processed/train_data_processed.csv
      - data/processed/validation_data_processed.csv

  model_training:
    cmd: python src/models/train_model.py
    deps:
      - src/models/train_model.py
      - params.yaml
      - data/processed/train_data_processed.csv
      - data/processed/validation_data_processed.csv
    outs:
      - models/model.pkl

  # evaluation:
  #   cmd: python evaluation.py
  #   deps:
  #     - evaluation.py
  #     - models/model.pkl
  #     - params.yaml
  #   outs:
  #     - metrics/metrics.json