Credit Risk Prediction

Overview

Credit risk assessment is crucial in the banking sector to minimize loan defaults and optimize lending strategies. This project develops a machine learning model to predict loan default probability based on applicant data.

Business Objective

• Identify key factors influencing loan defaults.
• Develop a predictive model for credit risk assessment.
• Provide actionable insights for financial institutions.

Dataset

The dataset consists of 45,000 loan applicants with 14 features, categorized as follows:

• Personal Information: Age, gender, education, income, employment experience, home ownership.
• Loan Details: Loan amount, intent, interest rate, loan-to-income ratio.
• Credit History: Credit score, credit history length, previous loan defaults.
• Target Variable: loan_status (1 = Default, 0 = Non-Default).

Data Preprocessing

• Missing Values: Handled appropriately to maintain data integrity.
• Categorical Encoding: Applied label encoding for categorical variables.
• Feature Scaling: Standardized numerical variables using StandardScaler.

Exploratory Data Analysis (EDA)

Key insights from data exploration:

• Income Distribution: Right-skewed, with a minority having high incomes.
• Credit Score vs Default: Higher credit scores correlate with lower default rates.
• Loan Amount vs Default: Larger loans increase default likelihood.
• Loan Purpose: Personal and business loans have higher default rates.

Machine Learning Model Development

Models Used

1. Random Forest Classifier
   • Accuracy: 85%, AUC-ROC: 0.78
   • Moderate performance, required recall improvement.
2. XGBoost Classifier (Best Model)
   • Accuracy: 89%, AUC-ROC: 0.84
   • Lower false positives, better classification of defaulters.

Feature Importance

Top predictors of loan default:

1. Credit Score – Most influential factor.
2. Loan Amount – Higher amounts correlate with higher risk.
3. Income-to-Loan Ratio – Lower ratios increase default probability.
4. Credit History Length – Longer histories reduce default likelihood.
5. Previous Defaults – Strong indicator of future default risk.

Business Recommendations

• Stronger Credit Score Thresholds: Enforce minimum credit score requirements.
• Income-Based Loan Caps: Limit loan amounts relative to income.
• Enhanced Risk Monitoring: Extra scrutiny for applicants with short credit histories.
• Risk-Based Interest Rates: Adjust rates for high-risk applicants to mitigate losses.