Refit a Pricing Model for an Insurance Comparison Site
Overview
What this challenge is about.
You receive 9 months of quote-impression data (about 14 million events) with about 60 features and click labels. Refit logistic regression with elastic-net regularization plus a structured interaction search (top-K interactions by mutual information). Compare against the production model on a temporal hold-out (last 30 days) using log-loss and calibration. Bootstrap confidence intervals. Deliver the modeling notebook, comparison report, and a 2-page memo with a ship/no-ship recommendation.
The Brief
What you'll do, and what you'll demonstrate.
Refit a pricing-relevant click model to improve log-loss without regressing calibration, and recommend ship or no-ship.
Earning criteria — what you'll demonstrate
- Apply elastic-net regularization to a real product model
- Search interactions in a principled (not ad-hoc) way
- Run temporal back-tests with bootstrap CIs
- Defend a ship/no-ship recommendation in writing
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Data Scientist
Refitting a production model with a defensible methodology and ship/no-ship memo is the classic senior data-science task in fintech.
This challenge sharpens
- regularized-regression
- feature-interactions
- calibration
Machine Learning Engineer
Temporal back-testing with bootstrap CIs is the same discipline MLEs apply to broader production-model rollouts.
This challenge sharpens
- model-comparison
- bootstrap-analysis
- calibration
Applied AI Scientist
Owning a model refit end-to-end including the leadership memo is the daily reality of applied-AI scientists in conversion-driven product orgs.
This challenge sharpens
- regularized-regression
- calibration
- model-comparison