Machine Learning-Based Loan Default Prediction System

Introduction

Welcome to the repository for my advanced machine learning project focused on predicting loan defaults. This project utilizes a robust, scalable architecture deployed on Microsoft Azure to offer precise default predictions, potentially saving financial institutions from significant losses due to risky exposures.

Project Overview

This project encompasses a full cycle of data handling, model development, and deployment, leveraging Azure's cloud capabilities to enhance predictive accuracy and operational efficiency.

Data Handling and Initial Analysis

• Data Preprocessing: Cleansing and structuring of dataset for optimal analysis.
• Exploratory Data Analysis (EDA): Identification of key predictors and understanding data patterns to set the stage for effective feature engineering.

Feature Engineering

Development of new features from existing data to improve the model's predictive accuracy regarding potential loan defaults.

Model Development and Selection

• Experimentation: Tested various models including Logistic Regression, Random Forest, and Gradient Boosting Machines.
• Model Selection: Chose the best-performing model based on accuracy, precision, recall, and F1-score.

Application Development and Modular Programming

Implemented modular programming to enhance maintainability and scalability of the application.

Developed a Flask web application for real-time data input and prediction output.

Deployment on Azure

• Containerization: Utilized Docker to create a consistent deployment environment.
• Azure Container Registry: Stored and managed Docker images.
• Continuous Deployment: Integrated with GitHub Actions for automated updates and deployment via Azure Web Apps.

Key Features

• Real-Time Predictions: Users can input data and receive predictions instantly.
• High Accuracy and Precision: Ensures reliable predictions to assist financial decision-making.
• Scalable Infrastructure: Built on Azure to efficiently handle increasing data and usage scales.

Technologies Used

• Microsoft Azure: For deploying and managing my application.
• Docker: For creating isolated environments to replicate production settings.
• Flask: For backend development of my web application.
• GitHub Actions: For CI/CD, ensuring seamless updates and deployments.

Getting Started

To get started with this project, please refer to the installation guide and follow the instructions to set up your environment.

Installation

Detailed steps to set up the project locally:

• Clone the repository:

git clone https://github.com/Abhi0323/Machine-Learning-Based-Loan-Default-Early-Warning-System

• Navigate to the project directory and install the dependencies:

cd Machine-Learning-Based-Loan-Default-Early-Warning-System
pip install -r requirements.txt

• Run the Flask application:

python app.py