Forecast Trust

Per-resort forecast model accuracy

Returns which weather model has been most accurate for a specific resort over the last 90 days (day-1 MAE, bias, sample count). Use when users ask how reliable the forecast is, which model to trust, or forecast verification for a named resort.

Status codes distinguish three cases that used to be conflated: an unknown resort is 404, a resort with no verification rows yet is 200 with zeroed stats and totalVerifications: 0, and a backend we cannot reach is 503. Treat 503 as retryable and 404 as terminal.

get/api/v1/resorts/{slug}/forecast-accuracy

Request

  • Base URL: https://www.snowsure.ai
  • URL: https://www.snowsure.ai/api/v1/resorts/{slug}/forecast-accuracy
  • Auth: none declared

Path parameters

slugstring required

Resort slug

Response

Per-model accuracy stats. A resort with no verification data yet returns 200 with zeroed stats and totalVerifications: 0 — not 404.

resortstring
resortNamestring
periodstring
bestModelstring nullable

Lowest day-1 error on snow-relevant days. Null when there were fewer than 10 snow-relevant days in the window (see snowRelevantDays) — too little snow to tell models apart — in which case models is empty.

recommendationstring

One sentence naming the best model, or saying there was not enough snow to rank models.

totalVerificationsinteger

Verified summit rows in the 90-day window — one per model, per forecast horizon, per day (~6 models x ~13 horizons a day). Not a day count; see verifiedDays.

verifiedDaysinteger

Distinct days in the 90-day window with at least one verified forecast.

snowRelevantDaysinteger

Days in the window where at least 1 cm of snow was observed or any model forecast at least 1 cm. Per-model error, bias and samples are measured over these days only; models are ranked once there are at least 10.

Example response

{
  "period": "90d"
}

Changes

No recorded changes to this operation across all 1 revision of this API.