How aitullalimon/end-to-end-ml-production-pipeline works
This repository is a reusable classification and regression ML pipeline template. Its documented workflow runs from data ingestion and EDA through preprocessing, training, validation, model saving, and batch inference. The notebook is the documented home for training; its implementation was not sampled, so training-stage links follow the README. The CLI’s model loading, prediction, and CSV export are supported by sampled source.
Main entry points: 01_setup_and_data.ipynb ML workflow · predict_cli.py Inference CLI · raw Raw datasets · data_loader.py Data loading
Components 14
Model development
- Data loading
src/data_loader.py - ML workflow
01_setup_and_data.ipynb - EDA reports
reports/eda - Preprocessing
01_setup_and_data.ipynb - Model training
01_setup_and_data.ipynb - Cross-validation
01_setup_and_data.ipynb - Evaluation metrics
01_setup_and_data.ipynb
Batch inference
- Saved models
- Inference CLI
src/predict_cli.py
Other components
- Notebook user
- Batch user
- Raw datasets
data/raw - Input CSV
data/sample_cls.csv - Predictions
Connections 12
- Notebook user → ML workflow: runs
- Raw datasets → Data loading: loads
- ML workflow → EDA reports: generates
- ML workflow → Preprocessing: prepares features
- Preprocessing → Model training: transforms
- Model training → Cross-validation: validates
- Model training → Evaluation metrics: evaluates
- ML workflow → Saved models: saves models
- Batch user → Inference CLI: runs
- Inference CLI → Input CSV: reads
- Inference CLI → Saved models: loads model
- Inference CLI → Predictions: writes CSV