Loan Approval Prediction

Project Overview

This project predicts whether a loan application will be approved or not using machine learning classification models. The dataset used is from Kaggle’s Loan Prediction problem. The goal is to build a robust model to assist banks or financial institutions in making automated loan approval decisions.

---

Tools & Libraries

• Python 3.x
• Pandas, NumPy for data processing
• Scikit-learn for model building and evaluation
• Seaborn, Matplotlib for visualization
• Joblib for model saving

---

Steps Performed

1. Data Loading & Exploration
   Explored dataset shape, features, missing values, and basic statistics.

2. Data Cleaning & Preprocessing

   • Imputed missing values with median (numerical) and mode (categorical).
   • Dropped irrelevant columns (Loan_ID).
   • Encoded categorical features using Label Encoding.
   • Scaled numerical features using StandardScaler.

3. Exploratory Data Analysis (EDA)
   Visualized distributions, relationships between features and target, and detected outliers.

4. Outlier Removal
   Removed outliers in ApplicantIncome and LoanAmount using the IQR method.

5. Feature Selection
   Selected relevant features based on correlation and feature importance from Random Forest.

6. Model Training
   Trained Logistic Regression, Random Forest, and SVM classifiers.

7. Hyperparameter Tuning
   Tuned Random Forest using RandomizedSearchCV for better performance.

8. Model Evaluation
   Evaluated models using accuracy, classification report.

9. Prediction on Test Data
   Processed test dataset similarly and predicted loan approval status.

10. Model Saving
    Saved the best model using Joblib for future inference.