MLproject

name: Cancer_Modeling

python_env: python_env.yaml

entry_points:
    data-preprocessing:
        parameters:
            test-size: {type: float, default: 0.33}  # Short syntax
            dvc-rev: {type: str, default: "v1.0"}
            dvc-message: {type: str, default: ""}
        command: "python data_preprocessing.py --test-size {test-size} --dvc-rev {dvc-rev} --dvc-message {dvc-message}"
    hyperparameters-tuning:
        parameters:
            data-run-id: {type: str, default: ""}
            data-version: {type: str, default: ""}
            n-trials:  # Long syntax
                type: int
                default: 10
        command: "python hyperparameters_tuning.py --data-run-id {data-run-id} --n-trials {n-trials} --data-version {data-version}"
    model-training:
        parameters:
            data-run-id: {type: str, default: ""}
            data-version: {type: str, default: ""}
            tuning-run-id: {type: str, default: ""}
        command: "python model_training.py --data-run-id {data-run-id} --tuning-run-id {tuning-run-id} --data-version {data-version}"
    model-evaluation:
        parameters:
            eval-dataset: str  # Specifying just the data type
        command: "python model_evaluation.py --eval-dataset {eval-dataset}"