dvc.yaml

stages:
  download_data:
    cmd: uv run predicting_outcomes_in_heart_failure/data/dataset.py
    deps:
      - predicting_outcomes_in_heart_failure/data/dataset.py
    outs:
      - data/raw/heart.csv

  preprocessing:
    cmd: uv run predicting_outcomes_in_heart_failure/data/preprocess.py
    deps:
      - predicting_outcomes_in_heart_failure/data/preprocess.py
      - data/raw/heart.csv
    outs:
      - data/interim/preprocessed.csv
      - data/interim/preprocessed_female_only.csv
      - data/interim/preprocessed_male_only.csv
      - data/interim/preprocessed_no_sex_column.csv
      - data/interim/preprocess_artifacts/scaler.joblib

  split_data:
    foreach: [all, female, male, nosex]
    do:
      cmd: >
        uv run predicting_outcomes_in_heart_failure/data/split_data.py
        --variant ${item}
      deps:
        - predicting_outcomes_in_heart_failure/data/split_data.py
        - data/interim/preprocessed.csv
        - data/interim/preprocessed_female_only.csv
        - data/interim/preprocessed_male_only.csv
        - data/interim/preprocessed_no_sex_column.csv
      outs:
        - data/processed/${item}

  training:
    foreach:
      - { variant: all,    model: logreg }
      - { variant: all,    model: random_forest }
      - { variant: all,    model: decision_tree }
      - { variant: female, model: logreg }
      - { variant: female, model: random_forest }
      - { variant: female, model: decision_tree }
      - { variant: male,   model: logreg }
      - { variant: male,   model: random_forest }
      - { variant: male,   model: decision_tree }
      - { variant: nosex,  model: logreg }
      - { variant: nosex,  model: random_forest }
      - { variant: nosex,  model: decision_tree }
    do:
      cmd: >
        uv run predicting_outcomes_in_heart_failure/modeling/train.py
        --variant ${item.variant}
        --model ${item.model}
      deps:
        - predicting_outcomes_in_heart_failure/modeling/train.py
        - data/processed/${item.variant}/train.csv
      outs:
        - models/${item.variant}/${item.model}.joblib
        - reports/${item.variant}/${item.model}

  evaluation:
    foreach:
      - { variant: all,    model: logreg }
      - { variant: all,    model: random_forest }
      - { variant: all,    model: decision_tree }
      - { variant: female, model: logreg }
      - { variant: female, model: random_forest }
      - { variant: female, model: decision_tree }
      - { variant: male,   model: logreg }
      - { variant: male,   model: random_forest }
      - { variant: male,   model: decision_tree }
      - { variant: nosex,  model: logreg }
      - { variant: nosex,  model: random_forest }
      - { variant: nosex,  model: decision_tree }
    do:
      cmd: >
        uv run predicting_outcomes_in_heart_failure/modeling/evaluate.py
        --variant ${item.variant}
        --model ${item.model}
      deps:
        - predicting_outcomes_in_heart_failure/modeling/evaluate.py
        - models/${item.variant}/${item.model}.joblib
        - data/processed/${item.variant}/test.csv
      outs:
        - metrics/test/${item.variant}/${item.model}.json