Mini-project
Insurance
Insurance claim fraud detection
Flag the few fraudulent claims among many honest ones, and explain each flag.
The brief
A small share of claims are fraudulent, and investigators cannot check every claim. Build a model that ranks claims by fraud risk and explains each flag, so investigators look at the right claims first and honest customers are paid quickly. Good looks like: metrics for an imbalanced problem, a threshold chosen by investigation capacity, and explanations investigators trust.
What you will build
- Explore the claims and the fraud label
- Engineer features from policy and claim details
- Train gradient boosting with imbalance handling
- Evaluate with precision at the investigators' capacity
- Explain each flag with SHAP
Free, open data to build it with
Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.