workshop-uncool-mlops/dvc.yaml

stages:

  get-data:
  
    cmd: python src/get_data.py
      --output_folder ${data.output_folder}

    deps:
      - src/get_data.py

    params:
      - data

    outs:
      - ${data.output_folder}

    metrics:
      - ${data.metrics_file}:
          cache: false

  split-data:
    cmd: 
      python src/split_data.py 
      ${data.output_folder} 
      ${split.output_folder} 
      ${split.test_size}
      ${split.metrics_file} 
    deps:
    - ${data.output_folder}
    - src/split_data.py
    outs:
    - ${split.output_folder}
    metrics:
    - ${split.metrics_file}:
        cache: false

  train:
    cmd: 
      python src/train.py
      ${split.output_folder}
      ${train.output_folder}
    deps:
    - ${split.output_folder}
    - src/train.py
    params:
      - data.labels
      - train
    outs:
    - ${train.output_folder}
    metrics:
      - ${train.metrics_folder}.json:
          cache: false
    plots:
      - ${train.metrics_folder}:
          cache: false

  eval:
    cmd: 
      python src/eval.py
      ${split.output_folder}/val.json
      ${train.output_folder}
      ${eval.output_folder}
    params:
      - data.labels
    deps:
    - ${split.output_folder}/val.json
    - ${train.output_folder}
    - src/eval.py
    - src/inference.py
    plots:
    - ${eval.output_folder}/plots/confusion_matrix.json:
        cache: false
        template: confusion
        x: actual
        y: predicted