Challenge: Classifier Model Feature Engineering

Apply what you’ve learned about feature engineering to predict the loan status in a coding exercise.

We'll cover the following

Problem statement

A financial institution has gathered data on various factors such as income, credit details, loan amount, term, and debt-to-income ratio from past applicants for Lending Club loans. It wants to use this data to predict the loan status of new applicants (i.e., whether they are likely to default). To achieve this, the financial institution built a classifier model using H2O GBM and achieved an AUC score of 0.90750.9075 on the test dataset. It has already tried hyperparameter tuning, and now it wants to further improve performance with feature engineering.

Your task is to create additional features to improve the test data auc score to >0.9085.> 0.9085. Try adding interaction terms, transforming variables, or creating new features based on domain knowledge.

Show your magic!

Get ready to improve the H2O GBM classifier model with your feature engineering techniques!

Click the “Run” button below to check out some dataset samples and the list of input features to help you think of generating additional features from them:

Get hands-on with 1200+ tech skills courses.