dvc.lock

schema: '2.0'
stages:
  get_data:
    cmd: python scripts/data.py
    deps:
    - path: scripts/data.py
      hash: md5
      md5: bb64dff68b4a519998e544185d7a49fe
      size: 1501
    params:
      params.yaml:
        index_col: flat_id
        table_name: real_estate_clean
    outs:
    - path: data/initial_data.csv
      hash: md5
      md5: 64da8c623e7bd964084ff1fefd9c70ed
      size: 14098801
      cloud:
        my_storage:
          etag: 64da8c623e7bd964084ff1fefd9c70ed
          version_id: 00061636A7CA030E
  fit_model:
    cmd: python scripts/fit.py
    deps:
    - path: data/initial_data.csv
      hash: md5
      md5: 64da8c623e7bd964084ff1fefd9c70ed
      size: 14098801
    - path: scripts/fit.py
      hash: md5
      md5: c27e06327c752f09d2afe0d3571bb1cb
      size: 2348
    params:
      params.yaml:
        cat_cols: []
        drop_cols:
        - id
        - building_id
        - flat_id
        one_hot_drop:
          binary: if_binary
          categorical: first
        target_col: price
    outs:
    - path: models/fitted_model.pkl
      hash: md5
      md5: 158aaec2832b5a9609e5f4032c694dcf
      size: 4718
      cloud:
        my_storage:
          etag: 158aaec2832b5a9609e5f4032c694dcf
          version_id: 00061636A7C29724
  evaluate_model:
    cmd: python scripts/evaluate.py
    deps:
    - path: models/fitted_model.pkl
      hash: md5
      md5: 158aaec2832b5a9609e5f4032c694dcf
      size: 4718
    - path: scripts/evaluate.py
      hash: md5
      md5: 9a6866c5fb18fde325b54d36e8ca782f
      size: 1345
    params:
      params.yaml:
        metrics:
        - neg_mean_squared_error
        - r2
        - neg_mean_absolute_percentage_error
        n_jobs: -1
        n_splits: 5
        target_col: price
    outs:
    - path: cv_results/cv_res.json
      hash: md5
      md5: 0e733ea51dc2cf8a34386d59296a2705
      size: 161
      cloud:
        my_storage:
          etag: 0e733ea51dc2cf8a34386d59296a2705
          version_id: 00061636A7BDE41A