dvc.yaml

stages:
  preprocess-data:
    cmd: poetry run bimodal preprocess-data
    deps: 
      - data/raw
      - python/src/bimodal/preprocessing.py
    outs:
      - data/counter_data.parquet
      - data/weather_rain.parquet
      - data/weather_wind.parquet
      - data/weather_temp.parquet
      - data/wellington_holidays.parquet

  build-db:
    cmd: poetry run bimodal build-db data/
    deps:
      - python/src/bimodal/database.py
      - data/counter_data.parquet
      - data/weather_rain.parquet
      - data/weather_wind.parquet
      - data/weather_temp.parquet
      - data/wellington_holidays.parquet
    outs:
      - data/db.sqlite

  build-datasets:
    cmd: poetry run bimodal build-datasets data/
    deps:
      - python/src/bimodal/datasets.py
      - data/counter_data.parquet
      - data/weather_rain.parquet
      - data/weather_wind.parquet
      - data/weather_temp.parquet
      - data/wellington_holidays.parquet
    outs:
      - data/dataset_daily_counts.parquet
      - data/dataset_hourly_counts.parquet