Predict Credit Default (Imbalanced)

Advanced classification project · 21 steps · Precision · recall

About

Binary credit outcomes are imbalanced: headline accuracy can hide poor detection of bad risk. The positive label is bad credit—optimize precision, recall, and threshold on probabilities, not accuracy alone. Uses a stratified train/test split.

Data

CSV at /datasets/german-credit/german.csv (German Credit data, OpenML-style labels). Source dataset is licensed under CC BY 4.0 with attribution to UCI.

Task

Complete TODOs in order. Use hints first. Switch to Sample solution when stuck.

Pro project

Guided projects are included with Pro. Upgrade to access all projects and run them in the browser.

Upgrade to Pro