Skip to main content
The forecast estimates what you’ll spend by the end of the current calendar month, in UTC, on all your spend or a filtered slice of it. Budgets use the same forecast over their own period. Read it with GET /api/v1/tenant/cost/forecast in the CostGraph API. It takes the same layer and filter parameters as cost queries, and works with an API key that has the focus:read scope.

​
The point and the range

The forecast returns one number and a range around it:
  • point is the best estimate of month-end spend: what you have already spent, plus charges already known for the rest of the month, plus a model of your usage for the remaining days.
  • lo and hi bound an 80% range. Month-end spend lands inside it about four months out of five.
actual is spend so far, knownFuture is charges already scheduled, and modelled is the estimate for the remaining days. The forecast is of gross spend, before credits.

​
How it forecasts

method says which model produced the forecast:
  • regression follows your recent trend and weekday pattern. CostGraph uses it once there are at least 14 days of history.
  • pace projects your average daily spend. CostGraph uses it with 7 to 13 days of history.

​
Minimum history

The forecast needs at least 7 days of spend history. With less, point, lo, and hi are empty and reason says why.

​
Currency

The forecast is in the currency most of the spend in scope is billed in. Spend in other currencies is listed in excludedCurrencies and left out. Filter by currency to forecast each one separately.