dvc.lock

schema: '2.0'
stages:
  raw_dataset_creation:
    cmd: python 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: 979221ddcebd0b130362f6bb96dc9caa
      size: 904
    outs:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 116146
  split_data:
    cmd: python 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: a98fe537e0ac4b6d2a2202a316200a3b
      size: 1361
    outs:
    - path: data/processed/churn_test.csv
      md5: c3a4965f71511e72d4af9ac650f47505
      size: 23358
    - path: data/processed/churn_train.csv
      md5: 8358e3ab11f8301c58d3b0e5aa99a6fd
      size: 92916
  model_train:
    cmd: python 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: d50f118f5b26792ac38c61d91ca89892
      size: 4112
    params:
      params.yaml:
        random_forest.max_depth: 100
        random_forest.n_estimators: 40
    outs:
    - path: report/metrics.json
      md5: 28f059fefa425d3e8e10fee318734056
      size: 143
    - path: report/params.json
      md5: ae7b42d948e61f866747b265c2a87659
      size: 48
  log_production_model:
    cmd: python src/models/production_model_selection.py --config=params.yaml
    deps:
    - path: src/models/production_model_selection.py
      md5: eff97533b415e7199a5ae4231509f2ce
      size: 1643
    params:
      params.yaml:
        random_forest.max_depth: 100
        random_forest.n_estimators: 40
    outs:
    - path: models/model.joblib
      md5: 35bf40c4e7d8f7abe7b291a8e1ab5795
      size: 651113