dvc.lock

schema: '2.0'
stages:
  data_ingestion:
    cmd: python src/data/data_ingestion.py
    deps:
    - path: src/data/data_ingestion.py
      hash: md5
      md5: 4d839b32b595c2a2e56acacabbd398b7
      size: 3562
    params:
      params.yaml:
        data_ingestion.test_size: 0.233
    outs:
    - path: data/raw
      hash: md5
      md5: 88fe84559a33b2c06347bfdc2cdf269a.dir
      size: 827973
      nfiles: 2
  data_preprocessing:
    cmd: python src/data/data_preprocessing.py
    deps:
    - path: data/raw
      hash: md5
      md5: 88fe84559a33b2c06347bfdc2cdf269a.dir
      size: 827973
      nfiles: 2
    - path: src/data/data_preprocessing.py
      hash: md5
      md5: c4e841a801f311a9f4906f07c6b0bcc2
      size: 4020
    outs:
    - path: data/interim
      hash: md5
      md5: 08118a82ddefa46014594d1bf23d70e7.dir
      size: 572486
      nfiles: 2
  feature_engineering:
    cmd: python src/features/feature_engineering.py
    deps:
    - path: data/interim
      hash: md5
      md5: 08118a82ddefa46014594d1bf23d70e7.dir
      size: 572486
      nfiles: 2
    - path: src/features/feature_engineering.py
      hash: md5
      md5: 5f38ede9fbeab0ff38341c158984b8e7
      size: 3885
    params:
      params.yaml:
        feature_engineering.max_features: 1000
    outs:
    - path: data/processed
      hash: md5
      md5: ca766ae74e185d6d36d24528f1ad9d46.dir
      size: 20786916
      nfiles: 2
    - path: models/vectorizer.pkl
      hash: md5
      md5: 2e5b958fc4610c9915ceff0ba809b679
      size: 159598
  model_building:
    cmd: python src/model/model_building.py
    deps:
    - path: data/processed
      hash: md5
      md5: ca766ae74e185d6d36d24528f1ad9d46.dir
      size: 20786916
      nfiles: 2
    - path: src/model/model_building.py
      hash: md5
      md5: fe1cb893fbd3f168fb5a1ca8eff480e3
      size: 2375
    outs:
    - path: models/model.pkl
      hash: md5
      md5: 070cd777a852b1f0a6035a42e797786d
      size: 8710
  model_evaluation:
    cmd: python src/model/model_evaluation.py
    deps:
    - path: models/model.pkl
      hash: md5
      md5: 070cd777a852b1f0a6035a42e797786d
      size: 8710
    - path: src/model/model_evaluation.py
      hash: md5
      md5: a145d2f40feb79af4ad4c9d13b58ebad
      size: 5888
    outs:
    - path: reports/experiment_info.json
      hash: md5
      md5: 7791eff894be18a85b0620bd023d1d6f
      size: 82
    - path: reports/metrics.json
      hash: md5
      md5: 8d153a34bb211cad9b4129e15983176b
      size: 145
  model_registration:
    cmd: python src/model/register_model.py
    deps:
    - path: reports/experiment_info.json
      hash: md5
      md5: 7791eff894be18a85b0620bd023d1d6f
      size: 82
    - path: src/model/register_model.py
      hash: md5
      md5: ba324c7a4487826eeb4ecf4c48478895
      size: 3100