dvc.yaml

stages:
  DATA_INGESTION:
    cmd: python src/stage_01_data_ingestion.py
    params:
      - params.py:
        - INPUT_FILE_NAME_REGEX
    deps:
      - src/stage_01_data_ingestion.py
      - src/utils/common.py
      - src/data_processing/data_validation.py
      - src/data_processing/data_transformation.py
      - src/data_processing/load_data.py
      - src/db_operation/db_operations.py
      - configs/config.py
    outs:
      - artifacts/sql_data_dir/Insurance_fraud_validated_data_db.db
      # - artifacts/local_data_dir/valid_data/model_training_validated_data_file.csv
      - artifacts/local_data_dir/Training_archive_bad_data

  DATA_PROCESSING:
    cmd: python src/stage_02_data_processing.py
    params:
      - params.py:
        - MAPPING_CATEGORICAL_COLUMNS
        - COLUMNS_TO_IGNORE_FOR_MODEL_TRAINING
    deps:
      - src/stage_02_data_processing.py
      - src/utils/common.py
      - configs/config.py
      - src/data_processing/data_preprocessing.py
      - src/db_operation/db_operations.py
      - artifacts/sql_data_dir/Insurance_fraud_validated_data_db.db
    outs:
      - artifacts/local_data_dir/processed_data/model_training_processed_data_file.csv

  DATA_CLUSTERING:
    cmd: python src/stage_03_data_clustering.py
    params:
      - params.py:
        - TARGET_COLUMN_NAME
        - CLUSTER_COLUMN_NAME
    deps:
      - src/stage_03_data_clustering.py
      - src/utils/common.py
      - configs/config.py
      - src/data_processing/data_clustering.py
      - artifacts/local_data_dir/processed_data/model_training_processed_data_file.csv

    outs:
      - artifacts/local_data_dir/clustered_data/model_training_clustered_data_file.csv
      - artifacts/graphs_dir/elbow_plot_cluster.png
      - artifacts/model_dir/data_clustering_model/k_means_model.h5
# Other stages...

  MODEL_TRAINING:
    cmd: python src/stage_04_model_training.py
    params:
      - params.py:
        - XGBOOST_HYPER_PARAMS
        - SVM_HYPER_PARAMS
    deps:
      - src/stage_03_data_clustering.py
      - src/stage_04_model_training.py
      - src/utils/common.py
      - configs/config.py
      - src/model_training/model_training.py
      - artifacts/local_data_dir/clustered_data/model_training_clustered_data_file.csv

    outs:
      - artifacts/model_dir/prediction_model_dir/