schema: '2.0'
stages:
  preprocess:
    cmd: python better_preprocess.py
    deps:
    - path: 25k_IMDb_movie_Dataset.csv
      hash: md5
      md5: 07989327c5c9a7d7a8d17e80e3e56fdf
      size: 12515047
    outs:
    - path: data.csv
      hash: md5
      md5: 0ae3ee0112169b038031c83f8e6e39c3
      size: 11917252
      isexec: true
  train:
    cmd: python script5_3.py