Bank loan prediction using  federation learning

The "Bank Loan Prediction using Federated Learning" project is designed to predict loan approval statuses based on personal and financial data. The project aims to provide accurate loan decisions by training machine learning models using a federated learning approach, ensuring privacy and efficiency. The dataset used consists of features such as age, gender, education, income, credit score, and loan details, with the target variable being loan approval status. Three machine learning algorithms CatBoost, XGBoost, and MLP are employed to build the predictive models. The project also includes an explainability module using LIME to interpret model predictions, enhancing transparency and trust in the system. The models are trained in a federated manner, where the data is divided across multiple clients, and their parameters are aggregated to create a global model. This approach ensures data privacy while maintaining predictive accuracy. The final model predicts whether a loan is approved or rejected based on the input features. The user-friendly web application built with Flask provides an interactive interface for users to input their data and view predictions along with explainability plots. This project provides a robust and private solution for loan approval prediction, combining advanced machine learning techniques and federated learning.

Keywords:

Loan approval, Federated learning, CatBoost, XGBoost, MLP, Predictive models, Data privacy, LIME, Machine learning, Flask.