Overview
What this challenge is about.
You receive 18,000 historical claims with text, attachments-count, claim amount, customer tenure, and the ground-truth final routing bucket. Train a structured classifier (e.g., a conditional random field over claim features, or a softmax classifier with a structured-output head) that respects the constraint that each claim must land in exactly one bucket. Compare against the existing 7-independent-binary-classifier baseline on macro-F1, routing consistency (no claim assigned to 2 buckets), and dollar-weighted error (a missed fraud-review on a EUR 80k claim hurts more than on a EUR 800 claim). Memo recommends keep, replace, or shadow-deploy.
The Brief
What you'll do, and what you'll demonstrate.
Replace 7 independent binary classifiers with a structured classifier so routing is consistent and dollar-weighted error drops.
Earning criteria — what you'll demonstrate
- Reframe a poorly-cast multi-binary problem as structured prediction
- Apply softmax or CRF heads to enforce output structure
- Design business-aligned evaluation metrics (dollar-weighted error)
- Plan a safe shadow deployment for a routing-critical model
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.
Machine Learning Engineer
Reframing a real production problem from independent binaries to structured prediction, with a deployment plan, is the senior-leaning skill MLE candidates are screened for.
This challenge sharpens
- structured-prediction
- multi-class-classification
- scikit-learn
Data Scientist
Designing dollar-weighted metrics and per-decile slices is the analytical depth that distinguishes a strong data scientist from a metrics-parrot.
This challenge sharpens
- model-evaluation
- feature-engineering
- structured-prediction