dvc.yaml

# implement dvc pipeline

stages:
  load_dataset:
    cmd: python src/data/load_dataset.py
    deps:
    - ./src/data/load_dataset.py
    params:
    - load_dataset.bucket
    - load_dataset.raw_data
    - load_dataset.file_name
    - load_dataset.aws_file_name
    outs:
    - .${load_dataset.raw_data}/${load_dataset.file_name}
  
  build_features:
    cmd: python src/features/build_features.py
    deps:
    - ./src/features/build_features.py
    - .${load_dataset.raw_data}/${load_dataset.file_name}
    params:
    - build_features.file_path
    - build_features.vectorizer_type
    - build_features.n_gram
    - build_features.max_features
    outs:
    - ./data/processed/clean_data.csv
  
  train_model:
    cmd: python src/models/train_model.py
    deps:
    - ./src/models/train_model.py
    - ./data/processed/clean_data.csv
    params:
    - build_features.vectorizer_type
    - build_features.n_gram
    - build_features.max_features
    - train_model.repo_owner
    - train_model.repo_name
    - train_model.model_name
    - train_model.experiment
    - train_model.experiment_name
    - train_model.experiment_description
    - train_model.hyperparams
    - train_model.test_size
    outs:
    - ./models/experiments/:
       persist: true
    - ./figures/training/:
       persist: true
    - ./vectorizer/:
       persist: true

  tune_model:
    cmd: python src/models/tune_model.py
    deps:
    - ./src/models/tune_model.py
    - ./data/processed/clean_data.csv
    params:
    - train_model.test_size
    - train_model.repo_name
    - train_model.repo_owner
    - tune_model.n_trials
    - tune_model.experiment_name
    - tune_model.experiment_description
    - tune_model.max_features
    - tune_model.ngram_range
    - tune_model.vectorizer_type
    outs:
    - ./models/fine_tunned/:
       persist: true
    - ./figures/tunning/