CodeAlpha_Credit-Scoring-Model

Machine Learning-based Credit Scoring System that predicts customer creditworthiness using Logistic Regression and Random Forest, achieving 79% accuracy through feature engineering and financial risk analysis.

Credit Scoring & Credit Risk Prediction using Machine Learning

Project Overview

This project develops a Machine Learning-based Credit Scoring System to evaluate the creditworthiness of loan applicants. By analyzing customer financial and demographic information, the model predicts whether an applicant represents a good or bad credit risk.

The goal is to support data-driven lending decisions, reduce the likelihood of loan defaults, and improve overall risk assessment.

---

Dataset Information

The dataset contains information related to loan applicants, including:

• Account status
• Credit history
• Loan purpose
• Credit amount
• Savings account status
• Employment duration
• Payment-to-income ratio
• Age
• Housing status
• Number of existing credits
• Guarantor information
• Collateral details

Target Variable:

• Good Credit Risk
• Bad Credit Risk

---

Exploratory Data Analysis (EDA)

Performed exploratory analysis to:

• Understand feature distributions
• Identify class imbalance
• Analyze relationships between variables
• Detect potential outliers and trends

---

Feature Engineering

To improve predictive performance, additional financial indicators were created:

Credit Per Month

credit_per_month = credit_amount / month_duration

Credit-Age Ratio

credit_age_ratio = credit_amount / age

Burden Score

burden_score = credit_amount * payment_to_income_ratio

These engineered features help capture an applicant's financial burden and repayment capability.

---

Data Preprocessing

• Label Encoding for ordinal categorical features
• One-Hot Encoding for nominal categorical features
• Feature Scaling using StandardScaler
• Train-Test Split for model evaluation

---

Machine Learning Models

The following classification algorithms were trained and evaluated:

Logistic Regression

A baseline linear classification model used for credit risk prediction.

Random Forest Classifier

An ensemble learning model that combines multiple decision trees to improve prediction accuracy and robustness.

---

Model Performance

Model	Accuracy
Logistic Regression	77%
Random Forest Classifier	78.5%

The Random Forest model achieved the best overall performance and was selected as the final model.

---

Technologies Used

• Python
• Pandas
• NumPy
• Matplotlib
• Scikit-learn

---

Project Workflow

1. Data Collection
2. Data Cleaning and Preprocessing
3. Exploratory Data Analysis
4. Feature Engineering
5. Model Training
6. Model Evaluation
7. Performance Comparison

---

Key Outcomes

• Built a Credit Risk Prediction System
• Applied Feature Engineering to financial data
• Compared multiple Machine Learning models
• Evaluated model performance using classification metrics
• Automated creditworthiness assessment

---

Future Improvements

• Hyperparameter Tuning using GridSearchCV
• Implementation of XGBoost and LightGBM
• Credit Score Generation (300–850 scale)
• Deployment using Streamlit or Flask
• Real-time Credit Risk Assessment Dashboard

---

Author

Janani M

Machine Learning | Data Science | Python