Loan Approval Prediction System

Problem Statement

Banks and financial institutions receive a large number of loan applications daily. Manually evaluating each application is time-consuming and can be prone to human bias or error. There is a need for a reliable, data-driven solution to automatically predict whether a loan application should be approved or not based on historical data and applicant details.

Project Goal

To develop a machine learning-based system that accurately predicts the approval status of a loan application using applicant-specific features such as income, education, and credit history. The system should be fast, scalable, and accessible through a RESTful API for easy integration into other applications or services.

Key Features

• 📊 Data Analysis & Visualization: Explored patterns in income, education, and credit history vs loan approval status using Seaborn and Matplotlib.

• 🔍 Data Preprocessing: Handled missing values, encoded categorical variables, and standardized numerical inputs.

• 🤖 Model Building & Evaluation:

  • Built multiple models: Logistic Regression, K-Nearest Neighbors (KNN), and Support Vector Machine (SVM).

  • Used GridSearchCV to tune hyperparameters and select the best-performing model.

• 🚀 API Integration: Deployed the final ML model via a FastAPI endpoint for real-time predictions.

• 💾 Model Serialization: Saved trained model and scaler using joblib for reuse in production.

Outcome

• Achieved ~85% accuracy using Logistic Regression and Support Vector Machine models.

• Built a working API where users can input five numeric features and receive instant loan approval predictions.

• Successfully deployed the ML model as a lightweight and scalable FastAPI microservice.

• This project demonstrates the ability to take a data science solution from EDA → Model Building → API Deployment.