Developer reports XGBoost forecasts county power outages two hours ahead
A developer built an hourly XGBoost system for 216 counties in Missouri and Illinois that estimates the probability of at least 5% of a county’s customers newly losing power during a six-hour window beginning two hours ahead. The model uses issued HRRR storm forecasts, county-specific information, current outages and neighboring-county data. It was trained on 2020–2024 and tested on 2025. The developer reported a PR-AUC of 0.094, compared with 0.039 for a rate-table baseline using National Weather Service warnings and Storm Prediction Center Day 1 outlooks. At equal alert volume, the developer reported catching 30% of new outages versus 17%, with 15% precision versus 11%. The base rate was 0.5% of county-hours. An early transfer test reportedly caught 31% versus 19% in Iowa, Indiana and Michigan.
The project reports a county-level test on 2025, a comparison with weather-warning data, paired storm-day bootstrap uncertainty estimates, a transfer test, and a live forecast whose outputs are being saved for a blind test. The developer also provided a write-up, code and a leakage audit. The reported results come from the developer.
Hi all! I built an hourly power outage forecast for 216 counties (MO + IL): P(≥5% of a county's customers newly out in the 6 h starting 2 h from now). Model: XGBoost on HRRR storm-related data (gusts, shear, CAPE, icing, etc.), county-specific facts, current outages and neighboring counties.
Setup decisions I'd like feedback on:
- Target is new outages only (no credit for predicting ones already underway). I thought predicting current outages was too easy and quite useless, so I chose to make it just for new outages, which was much harder to beat the ceiling.
- Trained 2020–2024, tested on 2025 using the HRRR forecasts as issued, the mean of the last 5 runs, not observed weather. The model validates on 2024, tests on 2025.
- Baseline: a rate table of NWS warnings + SPC Day 1 outlook, fit on the same rows.
- Uncertainty: bootstrap over storm days, paired.
Results on 2025: PR-AUC 0.094 vs 0.039 (ratio ~2.4×, 95% range 2.0–3.3×). At equal alert volume: 30% of new outages caught vs 17%; precision 15% vs 11%. Base rate is 0.5% of county-hours, so the precision is honest but low.
Early transfer test: trained only on MO + IL, it caught 31% vs 19% in Iowa, Indiana and Michigan (combined)
Live since Nov 1, with every forecast saved for a blind test. Write-up, code, and the leakage audit: https://ugobruzadin.github.io/storm-outage-risk/ code: https://github.com/UgoBruzadin/storm-outage-risk
#Storm Outage Risk Forecast#XGBoost#HRRR#National Weather Service#Storm Prediction Center