dvc.lock

schema: '2.0'
stages:
  data_preparation:
    cmd: python src/stage_01_prepare_data.py
    deps:
    - path: data/data.xml
      md5: a304afb96060aad90176268345e10355
      size: 37891850
    - path: src/stage_01_prepare_data.py
      md5: 0fb43c7f4f9018b69f9f31c54a1f8a23
      size: 2229
    - path: src/utils
      md5: eafd42bf6bdbf59daac33ed32def6db5.dir
      size: 9093
      nfiles: 9
    outs:
    - path: artifacts/prepared
      md5: dc639257fd7e43c167de7349d66763c4.dir
      size: 23886351
      nfiles: 2
  featurization:
    cmd: python src/stage_02_featurization.py
    deps:
    - path: artifacts/prepared
      md5: dc639257fd7e43c167de7349d66763c4.dir
      size: 23886351
      nfiles: 2
    - path: src/stage_02_featurization.py
      md5: e0b208949d523f376c550e5db13efb30
      size: 3266
    - path: src/utils
      md5: eafd42bf6bdbf59daac33ed32def6db5.dir
      size: 9093
      nfiles: 9
    outs:
    - path: artifacts/features
      md5: 2638a56d12c43583c71dc6f020a45b54.dir
      size: 12272480
      nfiles: 2
  train:
    cmd: python src/stage_03_train.py
    deps:
    - path: artifacts/features/train.pkl
      md5: 296954b676c7102e4212fa3e7d7691ed
      size: 8558640
    - path: src/stage_03_train.py
      md5: 77e0d974077c093b1dc633e3508363b4
      size: 2515
    - path: src/utils
      md5: eafd42bf6bdbf59daac33ed32def6db5.dir
      size: 9093
      nfiles: 9
    outs:
    - path: artifacts/model/model.pkl
      md5: 600f3ade64d84e757c4914a03f964fae
      size: 1718852
  evaluate:
    cmd: python src/stage_04_evaluate.py
    deps:
    - path: artifacts/features/test.pkl
      md5: 082d39185c64d407966f659a8f55a7fb
      size: 3713840
    - path: artifacts/model/model.pkl
      md5: 600f3ade64d84e757c4914a03f964fae
      size: 1718852
    - path: src/stage_04_evaluate.py
      md5: 62edef6cfd90002d61e8b490dc737472
      size: 3741
    - path: src/utils
      md5: eafd42bf6bdbf59daac33ed32def6db5.dir
      size: 9093
      nfiles: 9
    outs:
    - path: prc.json
      md5: b930642d7f036a3db97ee8b327473f79
      size: 136931
    - path: roc.json
      md5: b7978237f4e2558cb5c5c8ca0f7c8032
      size: 62037
    - path: scores.json
      md5: 27f74270394688e333d71191b17d0b9f
      size: 77