Skip to content

README.md

Latest commit

 

History

History
149 lines (102 loc) · 3.21 KB

README.md

File metadata and controls

149 lines (102 loc) · 3.21 KB
Raw

💰 Loan Prediction System

🔍 A Smart Loan Approval AI Tool Predict Loan Approval using Machine Learning , with REST API

Build Passing License: MIT Python Version Repo Size

📌 Overview

This project predicts Loan Approval with trained machine learning models such as (Logistic Regression , Ridge Classification, XGBOOST and Random forest Classifier )built on credit and income features. Additionally, it generates custom financial advice using a LLM, customized Report based on the outcome.

🔧 Stack:
FastAPI (backend) + SQLite3 (database) + React + Tailwind + TypeScript (frontend) + MLflow (tracking) + **LLM ** for insights.

🚀 Features

  • 📊 Trained ML model using feature engineering, PCA, and classification algorithms
  • 🧠 MLflow experiment tracking
  • 🔗 SQLite3 for relational data
  • 🧪 FastAPI backend for inference
  • 💻 React + TypeScript frontend with animated results
  • 💬 LLM response according to the Result
  • 🌐 Ready for cloud deployment

📦 Step 1: Environment Setup

✅ Python Version: 3.8 - 3.10 (Recommended < 3.11)

git clone https://github.com/Ajitkumar-1001/Loan_prediction_App.git
cd Loan_prediction_App
python -m venv Loan_App
source Loan_App/bin/activate
pip install -r requirements.txt

2️⃣ Database Setup

Create a SQLite database and required tables from the CSV dataset.

python create_database.py

3️⃣ Model Training + Logging

Train the model with feature selection, PCA, and classifiers like Logistic Regression, Random Forest, and Ridge. Log metrics using MLflow.

python train_model.py

Logged to dagshub

🧠 FastAPI Backend

🔧 Start the server:

uvicorn loan_api.main:app --reload

🔁 Endpoints:

  • POST /Loan/predict-loan → returns prediction + LLM-based suggestion

💻 React Frontend

📁 Navigate to frontend:

cd frontend
npm install

🚀 Start development server:

npm run dev
  • Form input fields
  • Animated prediction + suggestions
  • Validation to avoid empty or negative values

📬 API Example

POST /Loan/predict-loan
{
  "IncomePerDependent": 10000,
  "LoanAmount": 2500000,
  "RiskScore": 650,
  "TotalDebtToIncomeRatio": 0.4,
  "InterestRate": 7.5,
  "AnnualIncome": 60000,
  "BaseInterestRate": 5.5
}

Response:

{
  "prediction": "Approved",
  "message": "You are eligible for the loan.",
  "llm_response": "Based on your high income and low risk, we suggest going ahead with the loan application."
}

📚 Tech Stack

  • FastAPI
  • SQLite3
  • MLflow
  • React + TailwindCSS
  • LLM (Flash-2.0)
  • Docker-ready

🧑‍💻 Author

Ajit Kumar — GitHub

📜 License

This project is licensed under the MIT License.