Loan Prediction with AI and Streamlit

This project demonstrates how AI and Machine Learning can be applied to loan prediction in the financial industry. By leveraging Streamlit, we've deployed a Gaussian Naïve Bayes (GNB) model that predicts loan approval based on user inputs. The aim is to create a seamless, automated process for loan applications, improving efficiency and accuracy.

Demo

Try the loan prediction app here: Loan Checker AI

---

Features

• Automated Loan Predictions: Instantly predicts whether a loan will be approved or rejected based on user input.
• Streamlit Deployment: The app is deployed using Streamlit, offering an interactive and user-friendly experience.
• Gaussian Naïve Bayes Model: A machine learning model trained on a loan prediction dataset.
• Real-Time Predictions: The app processes user inputs in real-time to predict loan approval status.

Dataset

The dataset used in this project contains 13 variables, including:

• Loan_ID: Unique loan reference number
• ApplicantIncome: The income of the applicant
• LoanAmount: The loan amount requested
• Credit_History: Key variable indicating the applicant's credit history
• Property_Area: The area of the applicant's residence (Urban, Semi-Urban, or Rural)

Link to Dataset

Model Overview

The loan prediction model is built using Gaussian Naïve Bayes (GNB). Key performance metrics:

• Training Accuracy: 85.2%
• Test Accuracy: 81.82%

The model is trained to predict whether a loan will be approved based on various factors like credit history, income, and property area.

Installation

To run this project locally, follow these steps:

1. Clone the repository

git clone [https://github.com/githubaravind/loan-approval-prediction-using-AI]```

### 2. Install the required packages
Ensure you have Python 3.6+ and the following packages installed:
```bash
pip install streamlit pandas scikit-learn joblib

3. Run the application

Navigate to the project directory and run the app:

streamlit run main.py

The app will be available at http://localhost:8501.

Model Training

The model was trained on Kaggle using the loan prediction dataset. You can find the Kaggle notebook here:

• Kaggle Notebook

For training, we used the following features:

• Applicant Income
• Loan Amount
• Credit History
• Property Area
• Dependents

Exporting the Model

The trained model was exported using Pickle and Joblib for integration with the Streamlit app:

from joblib import dump
dump(GNB, 'gnb_model.joblib')

App Deployment

The app is deployed using Streamlit Community Cloud. To deploy the app yourself:

1. Push the repository to GitHub.
2. Log in to Streamlit.
3. Select your GitHub repository and deploy your app.

File Structure

Loan-Prediction/
│
├── .streamlit/
│   └── config.toml      # Streamlit configuration
├── dataset/
│   └── processed_dataset.csv  # Cleaned dataset used for prediction
├── models/
│   └── gnb_model.pkl    # Trained model
├── web/
│   ├── main.py          # Main application script
│   ├── predict.py       # Prediction logic
│   ├── dialogs.py       # Dialogs and UI elements
│   └── hooks.py         # Utility functions
└── README.md            # Project documentation

Contributing

Feel free to fork this repository, submit issues, or open pull requests. Contributions are welcome!

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Acknowledgments

• Kaggle Community for the loan prediction dataset and analysis tools.
• Streamlit for providing an easy-to-use deployment framework.
• Special thanks to Hassan for the insightful Exploratory Data Analysis (EDA) on the dataset.

---

If you found this project helpful or have any questions, feel free to reach out! reddyaravind2201@gmail.com