Accuracy report

This is the meter's first winter. After every storm that closes schools somewhere in the US we will post the numbers here: the chance the meter gave the evening before for a set of tracked districts, what those districts actually did, and the hit rate at each band (0–20 %, 20–40 %, and so on). A well-calibrated 70 % should be right about seven times in ten; that is the test.

What will be reported

• Storm date and region, the districts tracked, and the meter's 9 pm reading for each.
• Actual outcome: closed, delayed, e-learning, or open.
• Calibration by band and the biggest misses, with what the model got wrong (timing, ice, a district that never closes).
• Any change made to the model as a result, dated, so every reading can be reproduced.

Until the first storm

Read how the model works, run it on your town, and if it gets your district wrong this winter, email the date and what happened to the address on the about page. Every report goes into the calibration.

No storms scored yet · updated when the first one lands