Up to now, we have finished the process of model training, wepage development.

Model Training

We select the XGBoost and Logistic Regression for AAVE and Compound protocol, achieving accuracy of 90% and 80% respectively.

Webpage Development

We have built the webpage upon the Streamlit Framework for users to look up the credit rating results.

GitHub Link: https://github.com/sistine-yu/DeFiLend-Account-Risk.git