Incident severity predictor

The brief

An IT service desk handles thousands of incidents and only a few become major outages. The service manager wants a model that flags those few from the first details of a ticket, with a reason for each flag, so the right engineers are pulled in early. Good looks like: good recall on the rare major incidents, honest metrics for an imbalanced problem, and explanations an engineer agrees with.

What you will build

1. Explore the incident event log and define a major incident
2. Build features from the first events of each incident
3. Handle the imbalance and train gradient boosting
4. Evaluate with precision, recall and PR-AUC
5. Explain each flag with SHAP

Free, open data to build it with

• Incident management process enriched event log ↗ UCI Machine Learning Repository

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.