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: 09a1968ff3546a41aa9f69254c672e95
      size: 2013
    outs:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 120397
  split_data:
    cmd: python src/data/split_data.py --config=params.yaml
    deps:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 120397
    - path: src/data/split_data.py
      md5: c143a95ea697d55e3718f4fe9bdc4dfb
      size: 1383
    outs:
    - path: data/processed/churn_test.csv
      md5: c3a4965f71511e72d4af9ac650f47505
      size: 24209
    - path: data/processed/churn_train.csv
      md5: 8358e3ab11f8301c58d3b0e5aa99a6fd
      size: 96317
  model_train:
    cmd: python src/models/train_model.py --config=params.yaml
    deps:
    - path: data/processed/churn_test.csv
      md5: c3a4965f71511e72d4af9ac650f47505
      size: 24209
    - path: data/processed/churn_train.csv
      md5: 8358e3ab11f8301c58d3b0e5aa99a6fd
      size: 96317
    - path: src/models/train_model.py
      md5: 3d23b58cd5f1d6401a99ec42c6ebdb6e
      size: 3878
    params:
      params.yaml:
        random_forest.max_depth: 10
        random_forest.n_estimators: 50
  log_production_model:
    cmd: python src/models/production_model_selection.py --config=params.yaml
    deps:
    - path: src/models/production_model_selection.py
      md5: 00a34f68a8f67a05338190b2a8d079e3
      size: 1949
    params:
      params.yaml:
        random_forest.max_depth: 10
        random_forest.n_estimators: 50
    outs:
    - path: models/model.joblib
      md5: 6dd540667e5a15ee3324fc5756c900c0
      size: 1065406
  Monitor_Dashboard:
    cmd: python src/models/model_monitor.py --config=params.yaml
    deps:
    - path: data/raw/train_new.csv
      md5: b3058d1b696ea52f26871f1c8c71cb9c
      size: 23299
    - path: src/models/model_monitor.py
      md5: 4c0f6fdc7c2e4707fb1e7588ee7d1145
      size: 1541
    outs:
    - path: reports/data_and_target_drift_dashboard.html
      md5: 3f6616d53f55f7d882ac16566fd991a9
      size: 6300885
  great_expectation_validation:
    cmd: python src/data/data_validation.py --config=params.yaml
    deps:
    - path: data/raw/train.csv
      md5: 9b61c38a459d36244dbf8e6940236371
      size: 120397
    - path: data/raw/train_new.csv
      md5: b3058d1b696ea52f26871f1c8c71cb9c
      size: 23299
    - path: src/data/data_validation.py
      md5: b42ca893c03f7fcb25c9e3b8422eb100
      size: 3699
    outs:
    - path: great_expectations/exp_json_file.json
      md5: b1a5bd39b07da8f4d7be05576c3bd823
      size: 966