Predicted the customers’ probability of loan repayment on a joint dataset with loan history and credit card usage
• Predicted the customers’ probability of loan repayment on a joint dataset with loan history and credit card usage
• Compared models’ performance and chose gradient boosting tree because of its high accuracy of 73%
• Balanced dataset by oversampling and figured the best parameters for the model by performing grid search
• Visualized the feature engineering process using ggplot2 and demonstrated the result and analysis to our audience