Credit Card Default Prediction Classifier

Credit Card Default Risk Analysis

The purpose of this project is to conduct quantitative analysis on credit card default risk by using 3 machine learning models with accessible customer data, instead of credit score or credit history, with the goal of assisting and speeding up the human decision making process.

Dataset Source

This dataset contains information on default payments, demographic factors, credit limit, history of payments, and bill statements of credit card clients in Taiwan from April 2005 to September 2005.

Project Overview

The analysis consists of 2 Jupyter notebooks and a report explaining the overall workflow and methodology used:

1. Machine Learning Code
2. Credit Card Default Classifier Prediction Lieklihood Report

Machine Learning Models Used: 1.Logistic Regression 2.Linear Discriminant Analysis 3.Neural Network 4.Naive Bayes 5.K-Nearest Neighbours

Techniques Applied: 1.PCA 2.Oversampling (SMOTE) 3.Undersampling (randomsampler) 4.Combinedsampling (smote-enn)