Credit Card Default Classifier

A machine learning project for predicting credit card default using various classifiers and hyperparameter tuning techniques.

Project Overview

This project aims to predict credit card defaults using machine learning models. The dataset includes various features like customer information and financial behavior, which are used to predict the likelihood of a customer defaulting on a loan.

Machine Learning

Built machine learning pipelines using XGBoost, logistic regression, random forests, and boosted trees for credit default prediction

Feature Engineering

Engineered features to improve model performance, such as scaling, binning, and encoding categorical data

Model Evaluation

Evaluated models using F1-score and hyperparameter tuning to optimize performance

Technical Implementation

The project involved the following steps:

Data preprocessing, including handling missing values and scaling features
Comparison of multiple classifiers, including logistic regression, random forests, and XGBoost
Hyperparameter tuning to improve model performance and achieve optimal results
Use of cross-validation to prevent overfitting and assess generalization

Key Features Analyzed

Customer demographics (e.g., age, income, etc.)
Previous payment history and credit utilization
Loan status and default history

Project Results

0.535

F1-score

N/A

Precision

N/A

Recall

Key Findings

The project demonstrated an F1-score of 0.535, reflecting a balance between precision and recall for credit card default classification. The XGBoost model performed particularly well after hyperparameter tuning and feature engineering.

← Back to Portfolio