internship_project/dvc.yaml

stages: #stages of dvc
  #stage 1
  load_data: #specifying path running 
    cmd: python src/load_data.py --config=params.yaml #main .py path
    deps: #dependencis
    - src/get_data.py #location
    - src/load_data.py #location
    - data_given/insurance_updated.csv #root data location
    outs: #outcomes after running this code
    - data/raw/insurance_updated.csv
    #stage 2
  split_data:
    cmd: python src/split_data.py --config=params.yaml
    deps:
    - src/split_data.py
    - data/raw/insurance_updated.csv
    outs:
    - data/processed/train_insurance.csv
    - data/processed/test_insurance.csv 
    #stage 3
  train_evaluate:
    cmd: python src/train_evaluate.py --config=params.yaml
    deps:
    - data/processed/train_insurance.csv
    - data/processed/test_insurance.csv 
    - src/train_evaluate.py 
    params:
    - estimators.GradientBoostingRegressor.params.learning_rate
    - estimators.GradientBoostingRegressor.params.n_estimators
    - estimators.GradientBoostingRegressor.params.alpha
    - estimators.GradientBoostingRegressor.params.verbose
    - estimators.GradientBoostingRegressor.params.validation_fraction
    - estimators.GradientBoostingRegressor.params.tol
    - estimators.GradientBoostingRegressor.params.ccp_alpha
    metrics:
    - reports/scores.json:
        cache: false
    - reports/params.json:
        cache: false
    outs:
    - saved_models/model.pkl