dvc.lock

schema: '2.0'
stages:
  raw_dataset_creation:
    cmd: poetry run python3 src/data/load_data.py --config=params.yaml
    deps:
    - path: data/external/train.csv
      md5: 8929517566f004183dff480fdd48afe0
      size: 391872
    - path: src/data/load_data.py
      md5: 1de1b5780f0f897223d87024bae9c1fa
      size: 1392
    outs:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 116146
  split_data:
    cmd: poetry run python3 src/data/split_data.py --config=params.yaml
    deps:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 116146
    - path: src/data/split_data.py
      md5: c143a95ea697d55e3718f4fe9bdc4dfb
      size: 1354
    outs:
    - path: data/processed/churn_test.csv
      md5: c3a4965f71511e72d4af9ac650f47505
      size: 23358
    - path: data/processed/churn_train.csv
      md5: 8358e3ab11f8301c58d3b0e5aa99a6fd
      size: 92916
  model_train:
    cmd: poetry run python3 src/models/train_model.py --config=params.yaml
    deps:
    - path: data/processed/churn_test.csv
      md5: c3a4965f71511e72d4af9ac650f47505
      size: 23358
    - path: data/processed/churn_train.csv
      md5: 8358e3ab11f8301c58d3b0e5aa99a6fd
      size: 92916
    - path: src/models/train_model.py
      md5: afa9e12b72781911669685d91dfd02a6
      size: 3769
    params:
      params.yaml:
        random_forest.max_depth: 10
        random_forest.n_estimators: 29
  log_production_model:
    cmd: poetry run python3 src/models/production_model_selection.py --config=params.yaml
    deps:
    - path: src/models/production_model_selection.py
      md5: f77069f3ea8cd7e8a4eba646af470bfc
      size: 1668
    params:
      params.yaml:
        random_forest.max_depth: 10
        random_forest.n_estimators: 29
    outs:
    - path: models/model.joblib
      md5: 85ef87fbe017c2548065e6c9a6c9a044
      size: 599106