dvc.lock

schema: '2.0'
stages:
  data_split:
    cmd: python src/data_split.py
    deps:
    - path: src/data_split.py
      hash: md5
      md5: f1f19c4214eefec60fe741e7f8eafb8f
      size: 1327
    params:
      params.yaml:
        base.random_state: 47
        data_source.local_path: https://raw.githubusercontent.com/TripathiAshutosh/dataset/main/red_wine.csv
        split.split_ratio: 0.2
    outs:
    - path: data/split
      hash: md5
      md5: 0a00a1015edcc29ec1abc9043fac3ebb.dir
      size: 93746
      nfiles: 2
  data_processing:
    cmd: python src/data_processing.py data/processed
    deps:
    - path: data/split
      hash: md5
      md5: 0a00a1015edcc29ec1abc9043fac3ebb.dir
      size: 93746
      nfiles: 2
    - path: src/data_processing.py
      hash: md5
      md5: a3cebca64f34cdc6e32c02e77245768e
      size: 1449
    params:
      params.yaml:
        process.dir: data/processed
        process.test_file: final_test.csv
        process.train_file: final_train.csv
        split.dir: data/split
        split.test_file: test.csv
        split.train_file: train.csv
    outs:
    - path: data/processed
      hash: md5
      md5: d9b23e1f8a7adb4bd9679435aca9104c.dir
      size: 93746
      nfiles: 2
  train:
    cmd: python src/train.py data/features model/model.pkl
    deps:
    - path: data/processed
      hash: md5
      md5: d9b23e1f8a7adb4bd9679435aca9104c.dir
      size: 93746
      nfiles: 2
    - path: src/train.py
      hash: md5
      md5: 27f858660ca4698cdca67a4b64f43ba8
      size: 2481
    params:
      params.yaml:
        base.random_state: 47
        base.target_col: quality
        model_dir: model
        process.dir: data/processed
        process.test_file: final_test.csv
        process.train_file: final_train.csv
        train.n_est: 35
    outs:
    - path: model/model.pkl
      hash: md5
      md5: ec44c2b260d2ad0ea6570276dda712fc
      size: 591569
  evaluate:
    cmd: python src/evaluate.py model/model.pkl data/processed
    deps:
    - path: data/processed
      hash: md5
      md5: d9b23e1f8a7adb4bd9679435aca9104c.dir
      size: 93746
      nfiles: 2
    - path: model/model.pkl
      hash: md5
      md5: ec44c2b260d2ad0ea6570276dda712fc
      size: 591569
    - path: src/evaluate.py
      hash: md5
      md5: 576c42035c0bacac233c4b1737126894
      size: 6951
    outs:
    - path: eval/importance.png
      hash: md5
      md5: 77902a37686e49458e016017336ddf54
      size: 18510
    - path: eval/live/metrics.json
      hash: md5
      md5: d1d575afff31d3d6fd2b33627b6b0ccd
      size: 191
    - path: eval/live/plots
      hash: md5
      md5: a3f5f350054076942ec6fde0a11d391a.dir
      size: 112144
      nfiles: 4
    - path: eval/prc
      hash: md5
      md5: 891ecdb128e7b78a80fbd6803bd1c310.dir
      size: 9166
      nfiles: 2