Credit Card Default Analysis
Python
Machine Learning
Classification
Ensemble Methods
Explainability
Using XGBoosting to predict credit card defaults
Project Overview
A project of credit card default analysis and corresponding prediction.
https://github.com/SNALYF/Credit-Card-Default-Analysis
Key Concepts Applied
- EDA: Cleaned and implemented EDA to dataset, navigated on the correlations between numerical features and categorical features
- Model Selection: Conducted logistic model (Test F1-score: 0.530), SVC model (Test F1-score: 0.532), HistGradientBoosting (0.534), and XGBoosting model(0.507).
- Hyperparameter Optimization: Implemented Randomized Search Cross-Validation on SVC, HistGradientBoosting, and XGBoosting model to obtain the highest F1-score.
- Shap Interpretation: Analyzed SHAP plot to interpret black-box model decisions, identified most important feature.