🏦 Loan Default Prediction

A machine learning project to predict whether a client will default on a loan based on their financial and demographic data. Built for a hackathon using real-world NBFC (Non-Banking Financial Company) data.

---

🔍 Problem Statement

The goal is to develop a classification model that predicts the probability of a customer defaulting on a loan. This is a critical task for financial institutions to reduce risk and improve credit decision-making.

---

📊 Dataset Overview

The dataset includes features such as:

• Customer demographics (e.g., age, gender, education)
• Loan characteristics (e.g., loan amount, tenure, interest rate)
• Financial indicators (e.g., income, credit history)

⚠️ Dataset not included due to policy. If you are a recruiter or reviewer, please contact me or use synthetic data for testing.

---

🧠 Project Workflow

1. Data Preprocessing

   • Handling missing values
   • Label encoding / one-hot encoding
   • Scaling numerical features

2. Exploratory Data Analysis (EDA)

   • Distribution analysis
   • Correlation heatmaps
   • Class balance check

3. Model Building

   • XGBoost Classifier with tuning
   • 5-Fold Stratified Cross-Validation

4. Model Evaluation

   • Accuracy, Precision, Recall, F1-Score
   • ROC-AUC Score
   • Confusion Matrix

---

📈 Model Performance

Metric	Score
Cross-Validated Accuracy	XX%
ROC-AUC Score	XX%
Precision / Recall / F1	XX% / XX% / XX%

Final model: XGBClassifier(n_estimators=200, learning_rate=0.05, max_depth=4)

---

📦 Tech Stack

• Python 3.x
• Pandas, NumPy
• Scikit-learn
• XGBoost
• Matplotlib / Seaborn

---

🚀 Future Improvements

• Address class imbalance (SMOTE, ADASYN, or scale_pos_weight)
• Hyperparameter tuning (GridSearch / Optuna)
• Feature importance analysis
• Streamlit dashboard for predictions

---

🧾 License

This project is open-source under the MIT License.

---

🙋‍♂️ Author

Ankush Waghmare 📧 ankush.waghmare3804@gmail.com