AI-Based Decision Support System for Loan Risk Prediction

Project Overview

This project develops an AI-based decision support system for loan risk prediction using machine learning techniques. The system is designed to support financial decision-making by combining prediction performance with interpretable outputs.

The application compares multiple machine learning models and provides practical decision-support outputs, including predicted loan decisions, approval probabilities, suggested lending actions, and key decision factors.

Objectives

The main objectives of this project are:

to develop a machine learning-based prototype for loan risk prediction
to compare multiple classification models
to evaluate the system in terms of predictive performance, interpretability, and usability
to demonstrate how AI can support financial decision-making in a practical way

Models Used

The system compares the following machine learning models:

Logistic Regression
Decision Tree
Random Forest
Gradient Boosting

Main Features

Upload training CSV file
Upload testing CSV file
Automatic data preprocessing
Model training and comparison
Confusion matrix display
Manual single-row prediction
Approval probability output
Suggested lending action
Key decision factor explanation
Download prediction results as CSV

Dataset

This project uses the publicly available Loan Prediction dataset from Kaggle.

Training data includes the target variable:

Loan_Status

Example input variables include:

ApplicantIncome
CoapplicantIncome
LoanAmount
Loan_Amount_Term
Credit_History
Gender
Married
Dependents
Education
Self_Employed
Property_Area

Technologies

Python
Streamlit
pandas
numpy
scikit-learn

How to Run

Open a terminal in the project folder and run:

streamlit run loan_risk_app.py

If needed, you can also run:

python -m streamlit run loan_risk_app.py

fengx29/Xingkai_Feng_16038226_6069CEM_Individual_Project

About

Resources

Stars

Watchers

Forks

Releases

Languages