kenghooi-teoh/skyteam-zenml

Predicting loan default

Project summary

Using Zenml to create model training and inference pipelines to perform loan default prediction. Check here for a longer writeup.

Python Requirements

pip install -r requirements.txt

Initialize .zen environment

zenml init

Setup stack

bash stack-setup.sh

Run app using launch.sh

There bash script does the following:

1. Start a MySQL database container
2. Populate the database
3. Start the Streamlit app, which can be accessed at http://localhost:8501

Project tree:

📦 
├─ .gitignore
├─ README.md
├─ app
│  ├─ Home.py
│  ├─ __init__.py
│  ├─ drift_utils.py
│  ├─ mlpipeline
│  │  ├─ __init__.py
│  │  ├─ config.py
│  │  ├─ pipelines
│  │  │  ├─ __init__.py
│  │  │  ├─ batch_inference_pipeline.py
│  │  │  ├─ monitoring_pipeline.py
│  │  │  ├─ single_inference_pipeline.py
│  │  │  ├─ training_pipeline.py
│  │  │  └─ utils.py
│  │  ├─ run_batch_inference_pipeline.py
│  │  ├─ run_jz_training_pipeline.py
│  │  ├─ run_retraining_pipeline.py
│  │  ├─ run_single_inference_pipeline.py
│  │  ├─ run_training_pipeline.py
│  │  ├─ steps
│  │  │  ├─ __init__.py
│  │  │  ├─ data_fetcher.py
│  │  │  ├─ data_preprocessor.py
│  │  │  ├─ feature_engineer.py
│  │  │  ├─ model_evaluator.py
│  │  │  ├─ prediction_service_loader.py
│  │  │  ├─ prediction_storer.py
│  │  │  ├─ predictor.py
│  │  │  ├─ trainer.py
│  │  │  ├─ training_config.py
│  │  │  └─ util.py
│  │  └─ tests
│  │     ├─ __init__.py
│  │     └─ data_test.py
│  ├─ pages
│  │  ├─ 01_Inference.py
│  │  ├─ 02_Default_report.py
│  │  ├─ 02_data_drift.py
│  │  └─ 03_Data_drift.py
│  └─ st_utils.py
├─ data-ingestion
│  └─ load_data.py
├─ deploy.sh
├─ docs
│  ├─ _config.yml
│  └─ blog.md
├─ notebook
│  ├─ data-split.ipynb
│  ├─ test_sql.ipynb
│  └─ xgboost.ipynb
├─ requirements.txt
└─ stack-setup.sh

©generated by Project Tree Generator