dvc.yaml

stages:
  data_split: 
    cmd: python stages/data_split.py --config=params.yaml
    deps:
      - stages/data_split.py
      - ${base.data_dir}/${base.data_fname}
    params:
      - base
      - data_split
    outs:
      - data/X_train.pkl
      - data/X_test.pkl
      - data/y_train.pkl
      - data/y_test.pkl
  train:
    cmd: python stages/train.py --config=params.yaml
    deps:
      - stages/train.py
      - data/X_train.pkl
      - data/y_train.pkl
    params:
      - base
      - train
    outs:
      - ${base.model_dir}/clf-model
      - ${base.model_dir}/clf-model.mlem:
          cache: false
  eval:
    cmd: python stages/eval.py --config=params.yaml
    deps:
      - stages/eval.py
      - data/X_test.pkl
      - data/y_test.pkl
      - ${base.model_dir}/clf-model
      - ${base.model_dir}/clf-model.mlem
    params:
      - base
      - data_split
      - train
    outs:
      - ${eval.perm_imp_model_path}
      - feat_imp.csv:
          cache: false
    metrics:
      - metrics.json:
          cache: false
    plots:
      - eval_plots/cm.png:
          cache: false
      - eval_plots/plots/sklearn/roc.json:
          cache: false
          x: fpr
          y: tpr