stages:
  prepare:
    cmd: python3 process_dataset.py
    deps:
    - lego_sets.csv
    - process_dataset.py
    outs:
    - lego_sets_clean.csv
    - lego_sets_clean_train.csv
    - lego_sets_clean_valid.csv
    - lego_sets_clean_test.csv
  train:
    cmd: python3 simple_regression.py 100
    deps:
    - lego_sets_clean_train.csv
    - lego_sets_clean_test.csv
    outs:
    - lego_reg_model
    - lego_reg_results.csv