dvc.yaml

vars:
- PIPELINE_DIR: pipelines/train

stages:

  load_data:
    cmd: |
      echo "Download Bike Sharing data..."
      wget https://archive.ics.uci.edu/ml/machine-learning-databases/00275/Bike-Sharing-Dataset.zip -P data/
      echo "Unzip data..."
      unzip data/Bike-Sharing-Dataset.zip -d data/
    wdir: ../..
    outs:
    - ${data.raw_data}:
        cache: false

  extract_data:
    cmd: python src/stages/extract_data.py --config=${PIPELINE_DIR}/params.yaml
    wdir: ../..
    params:
    - ${PIPELINE_DIR}/params.yaml:
      - extract_data
    deps:
    - src/stages/extract_data.py
    - ${data.raw_data}
    outs:
    - ${data.train_data}
    - ${data.test_data}

  train:
    cmd: python src/stages/train.py --config=${PIPELINE_DIR}/params.yaml
    wdir: ../..
    params:
    - ${PIPELINE_DIR}/params.yaml:
      - train
    deps:
    - src/stages/train.py
    - ${data.train_data}
    outs:
    - ${train.model_path}

  evaluate:
    cmd: python src/stages/evaluate.py --config=${PIPELINE_DIR}/params.yaml --pdir=${PIPELINE_DIR}
    wdir: ../..
    params:
    - ${PIPELINE_DIR}/params.yaml:
      - evaluate
    deps:
    - src/stages/evaluate.py
    - ${data.test_data}
    - ${data.train_data}
    - ${train.model_path}
    outs:
    - ${data.reference_data}
metrics:
- ../../reports/train/metrics.json