INTELLIGENT CREDIT RISK SCORING & AGENTIC LENDING DECISION SUPPORT

Project Overview

This project involves the design and implementation of an system that evaluates borrower credit risk and evolves in decision support assistant. AI-driven credit analytics of an agentic AI lending.

---

Problem Statement

Financial institutions face significant challenges in assessing borrower creditworthiness accurately. Manual risk evaluation processes are often time consuming, inconsistent, and prone to human bias.

This project addresses this problem by implementing an automated credit risk scoring system that uses machine learning algorithms to analyze borrower data and classify applicants into risk categories.

Key Features

• Upload borrower dataset through an interactive UI
• Automatic data preprocessing pipeline
• Support for categorical encoding and feature scaling
• Training and comparison of multiple ML models
• Real-time credit risk prediction
• Visualization of evaluation metrics
• Clean and user-friendly Streamlit interface

Machine Learning Models Used

The following supervised learning models were implemented:

Logistic Regression

• Used for probabilistic classification
• Estimates default likelihood

Decision Tree Classifier

• Rule-based classification model
• Identifies important risk driving features

Evaluation Metrics

Model performance is evaluated using:

• Accuracy Score
• ROC-AUC Score
• Confusion Matrix
• ROC Curve Visualization
• Feature Importance Analysis

Installation and Setup Instructions

Follow these steps to run the project locally.

Step 1: Clone the Repository

git clone https://github.com/CWAbhi/Gen-AI_Capstone.git
cd Gen-AI_Capstone

Step 2: Install Dependencies

pip install -r requirements.txt

Step 3: Launch the Application

Start the Streamlit server:

streamlit run app.py

The application will open automatically in your browser.

Team Contribution

Member	Contribution
Anshika Seth (2401010080)	Data Cleaning & EDA, Complete Model Development, Streamlit UI, Deployment
Abhijeet Dey (2401010014)	Helped Model Development, Deployment
Aditya Ranjan (2401010035)	Documentation & Testing

Conclusion

The Credit Risk Prediction System successfully demonstrates how Machine Learning can automate loan risk assessment. The trained model achieved strong performance and can assist financial institutions in making reliable lending decisions.