Fetal-Health/dvc.yaml

stages:
  data_load:
    cmd: python -m stages.data_load
    deps:
    - stages/data_load.py
    # params:
    # - base
    # - data_load
    outs:
    - data/raw/fetal_health.csv

  data_split:
    cmd: python -m stages.data_split
    deps:
    - data/raw/fetal_health.csv
    - stages/data_split.py
    # params:
    # - base
    # - data_split
    # - featurize
    outs:
    - data/processed/test.csv
    - data/processed/train.csv

  featurize:
    cmd: python -m stages.featurize 
    deps:
    - data/processed/test.csv
    - data/processed/train.csv
    - stages/featurize.py
    # params:
    # - base
    # - data_load
    # - featurize
    outs:
    - data/processed/featured_train.csv
    - data/processed/featured_test.csv

  train:
    cmd: python -m stages.train
    deps:
    - data/processed/featured_train.csv
    - stages/train.py
    # params:
    # - base
    # - data_split.trainset_path
    # - featurize.target_column
    # - train
    outs:
    - models/model.joblib

  evaluate:
    cmd: python -m stages.evaluate
    deps:
    - data/processed/featured_test.csv
    - models/model.joblib
    - stages/evaluate.py
    # params:
    # - base
    # - data_split.testset_path
    # - evaluate
    # - featurize.target_column
    outs:
    - reports/confusion_matrix.png
    plots:
    - reports/confusion_matrix_data.csv:
        template: confusion
        x: predicted
        y: y_true
    metrics:
    - reports/metrics.json:
        cache: false