Home Credit Default Risk (Group)

Case synopsis

Home Credit cares deeply about the population without sufficient credit history and aims to improve inclusion for this underrepresented group by designing prediction models for loan repayment, such that capable borrowers are not denied solely due to absence of credit while protecting against defaults.

Group Members

Georgia Christodoulou

Andy Spendlove

Adam Bushman

Final Model

Fitting and predicting "default" with final model selection.

Process Summary

Summarizing the process of model exploration.

Model Performance Results

Visualizing the results of model exploration.

Model: Random Forest

Exploring the performace of a random forest model on predicting "default".

Model: Support Vector Machine

Exploring the performace of a support vector machine model on predicting "default".

Model: Penalized Regression

Exploring the performace of an elastic net, penalized logistic regression model on predicting "default".

Majority Class Classifier

Generating a baseline of performance for predicting "default" using the majority class classifier.

Data Preparation

Pipeline for data wrangling, cleaning, and class balancing of the source Home Credit data.