Overview
What this challenge is about.
You receive 12 months of 10-minute SCADA data (wind speed, air temperature, power output) for 30 representative turbines, plus the manufacturer's published curve. Fit a GP with an RBF + Matern52 kernel on each turbine's daily resampled data, holding out the last 6 weeks for validation. Flag a turbine as underperforming when more than 30 percent of recent observations fall below the GP's 90 percent lower-credible-bound for their wind speed. Validate against a list of 6 turbines the asset team already knows are underperforming. Success is recall of at least 5/6 known cases with no more than 4 false positives across the 30 turbines.
The Brief
What you'll do, and what you'll demonstrate.
Use per-turbine Gaussian Process regression to flag genuine underperformance with calibrated uncertainty and a defensible false-positive budget.
Earning criteria — what you'll demonstrate
- Apply Gaussian Process regression with composed kernels to real time-series data
- Tune kernel hyperparameters via marginal likelihood maximization
- Use credible bounds to define a defensible anomaly threshold
- Communicate GP-based decisions to a non-statistician asset team
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.
Career paths this builds toward
Canonical rolesData Scientist
Probabilistic regression with calibrated bounds on industrial sensor data is the daily work of data scientists at energy and infrastructure firms.
This challenge sharpens
- gaussian-processes
- uncertainty-quantification
- anomaly-detection
Applied AI Scientist
Choosing kernels, validating coverage, and translating credible bounds into a flagging threshold is the rhythm of applied AI in operational settings.
This challenge sharpens
- gaussian-processes
- kernel-methods
- uncertainty-quantification
Machine Learning Engineer
Productionizing a per-asset GP pipeline with reproducible artifacts and a flagging report is core MLE work in industrial AI.
This challenge sharpens
- python
- kernel-methods
- anomaly-detection