End to End Data Science Project — Wine Quality

End-to-end ML pipeline for predicting wine quality, with automated ingestion → validation → transformation → training → evaluation, plus a simple Flask UI for inference.

Overview

• Orchestration: main.py runs all stages sequentially.
• Serving: app.py exposes /train (runs main.py) and /predict (UI form → prediction).
• Model: ElasticNet regression with tunable alpha and l1_ratio from params.yaml.
• Tracking: MLflow logging and optional registry via src.datascience.components.model_evaluation.ModelEvaluation.log_into_mlflow.

Project Structure

.
├─ main.py
├─ app.py
├─ config/
│  └─ config.yaml
├─ params.yaml
├─ schema.yaml
├─ src/datascience/
│  ├─ constants/__init__.py
│  ├─ config/configuration.py
│  ├─ entity/config_entity.py
│  ├─ utils/common.py
│  ├─ components/
│  │  ├─ data_ingestion.py
│  │  ├─ data_validation.py
│  │  ├─ data_transformation.py
│  │  ├─ model_trainer.py
│  │  └─ model_evaluation.py
│  └─ pipeline/
│     ├─ data_ingestion_pipeline.py
│     ├─ data_validation_pipeline.py
│     ├─ data_transformation_pipeline.py
│     ├─ model_trainer_pipeline.py
│     ├─ model_evaluation_pipeline.py
│     └─ prediction_pipeline.py
├─ templates/
│  ├─ index.html
│  └─ results.html
└─ artifacts/ (generated)

Configuration

• Artifacts and paths: config/config.yaml
• Schema (columns, TARGET_COLUMN): schema.yaml
• Hyperparameters: params.yaml
• Constants: src.datascience.constants
• Config loader: src.datascience.config.configuration.ConfigurationManager

Pipelines

• Ingestion: src.datascience.pipeline.data_ingestion_pipeline.DataIngestionTrainingPipeline
• Validation: src.datascience.pipeline.data_validation_pipeline.DataValidationTrainingPipeline
• Transformation: src.datascience.pipeline.data_transformation_pipeline.DataTransformationTrainingPipeline
• Training: src.datascience.pipeline.model_trainer_pipeline.ModelTrainerTrainingPipeline
• Evaluation/MLflow: src.datascience.pipeline.model_evaluation_pipeline.ModelEvaluationTrainingPipeline
• Prediction (serving): src.datascience.pipeline.prediction_pipeline.PredictionPipeline

Setup

python -m venv env
source env/bin/activate
pip install -r requirements.txt
cp .env.example .env  # optional, for MLflow credentials if needed

How to Run (Training)

python main.py

Artifacts are written under artifacts/:

• Raw CSV: artifacts/data_ingestion/winequality-red.csv
• Splits: artifacts/data_transformation/train.csv, artifacts/data_transformation/test.csv
• Model: artifacts/model_trainer/model.joblib
• Metrics: artifacts/model_evaluation/metrics.json
• Logs: logs/logging.log

How to Run (Web UI)

python app.py  # serves on http://0.0.0.0:8080

• / renders templates/index.html (feature form).
• /predict returns prediction via templates/results.html.
• /train triggers full pipeline via main.py.

Guidelines for Extending

1. Update schemas in schema.yaml when adding/removing columns.
2. Adjust paths/artifacts in config/config.yaml when changing storage layout.
3. Tune hyperparameters in params.yaml.
4. Add components under src/datascience/components and wire them in src.datascience.config.configuration.ConfigurationManager and pipelines under src/datascience/pipeline.
5. Keep utility helpers in src/datascience/utils/common.py for I/O and serialization.
6. Ensure MLflow URI is set in src.datascience.config.configuration.ConfigurationManager.get_model_evaluation_config if remote tracking is used.

License

This project is licensed under GPL-3.0; see LICENSE.