dvc.yaml

stages:
    data_ingestion:
            cmd: python src/pipeline/stage_01_data_ingestion.py
            deps:
                - src/pipeline/stage_01_data_ingestion.py
                - params.yaml
            params:
                - data_ingestion.root_dir
                - data_ingestion.source_URL
                - data_ingestion.local_data_file
                - data_ingestion.unzip_dir
            outs:
                - artifacts/data_ingestion/data/
                - artifacts/data_ingestion/ingected_data/ingected_test_dir
                - artifacts/data_ingestion/ingected_data/ingected_train_dir

    data_validation:
        cmd: python src/pipeline/stage_02_data_validation.py
        deps:
            - src/pipeline/stage_02_data_validation.py
            - src/components/data_validation.py
            - params.yaml
        params:
            - data_validation.root_dir
            - data_validation.schema_path_file
            - data_validation.report_file_path
            - data_validation.report_page_file_path
        outs:
            - artifacts/data_validation/  
    data_preprocessing:
        cmd: python src/pipeline/stage_03_data_preprocessing.py
        deps:
            - src/pipeline/stage_03_data_preprocessing.py
            - src/components/data_preprocessing.py
            - params.yaml
        params:
            - data_preprocessing.root_dir
            - data_preprocessing.elbow_file_path
            - data_preprocessing.cluster_number_path
            - data_preprocessing.cluster_model_path
        outs:
            - artifacts/data_preprocessing