AppliedMultiple choice
Your churn model’s most predictive feature is "number of support calls in the thirty days before the account closed". Validation accuracy is superb. What is wrong?
- ANothing; support calls are a genuine early warning sign of churn
- BIt is leakage: the window is anchored to the closure date
- CThe feature should be standardised before the accuracy can be trusted
- DTree models cannot use count features without binning them first
The worked solution is in Premium
The answer, the full working and the one idea to take away – for this and all 1,322 questions in the bank. Answer it in practice and your working is marked, with a known mistake named when you make one.
Learn the method
More research case studies questions
- A model’s errors on five cases are 1,−1,2,−2,6.Foundation
- Pricing houses in logs, part 1 of 3Foundation
- Your churn model feeds a retention offer with a fixed budget.Applied
- You are asked how to validate a demand forecasting model built on three years…Applied
- An opera house expects ticket demand q = 1000−10p at price p and has 800 seats.Applied
- A customer churns with probability 0.3 and is worth $200 if retained.Applied