schema: '2.0'
stages:
  prepare:
    cmd: python create_data.py
    deps:
    - path: KaggleV2-May-2016.csv
      md5: cc55525de6e2b615aeba50095e8aaa95
      size: 10850063
    - path: create_data.py
      md5: 0cefc3631ca8d15d62df548c92d53eb3
      size: 2437
    outs:
    - path: test_dataset_dvc.csv
      md5: b0d2e25243ff9f564546becf8464af55
      size: 836063
    - path: train_dataset_dvc.csv
      md5: a17d28d659e1ba5f62a4203b7635ccfe
      size: 3345298
  train:
    cmd: python train_model.py
    deps:
    - path: test_dataset_dvc.csv
      md5: b0d2e25243ff9f564546becf8464af55
      size: 836063
    - path: train_dataset_dvc.csv
      md5: a17d28d659e1ba5f62a4203b7635ccfe
      size: 3345298
    outs:
    - path: logs_dvc.txt
      md5: 506eda87493d4f7cc0ba2b7389b5a182
      size: 52