API Reference
Perform forecasts with foundation models over a timeseries family.
Run a forecast query
Perform forecasts with foundation models. The input series may first be aggregated across dimensions and hierarchies.
query Parameters
modelThe forecasting model to use, e.g. t0-alpha.
Headers
AuthorizationThe Authorization header is used to authenticate with the API using your API key. Value is of the format Bearer YOUR_KEY_HERE.
Run a forecast query › Request Body
A timeseries family written out in full: where its rows come from, plus what its columns mean.
targetsThe columns to forecast. Omit to forecast every column the schema
declares with the "target" kind.
Selector for materializing a subset of a timeseries family. If omitted, every timeseries will be forecasted.
An object with one entry per dimension, keyed by the level to forecast that
dimension at: "city" takes the geography dimension at its city level,
aggregating the finer levels below it. Each value is a pattern, narrowing
which of that level's values to take:
"any"selects every value at that level.[value, ...]selects concrete values.{ "from": value?, "to": value? }selects a bounded or half-bounded range.
A dimension with no entry is aggregated over entirely.
{}quantilesQuantile levels to forecast for every target, as numbers in
[0, 1]; include "mean" to also request the mean. Omit for a
default spread.
compute_metricsScore the forecast against the actuals the request carries, and
report the scores under metrics in the response. A cutoff is
scored only when its whole forecast window is present and non-null
for every target. Some of the scores are computed on the mean, so
when this is set the mean is forecasted whether or not quantiles
asks for it.
A non-empty array of information cutoffs. For example ["latest"], an
explicit date ["2025-01-31"], or a recurring schedule [{ "every": "3d" }].
["latest"]Gap held before the forecast starts, on top of the step separating the information cutoff from the first prediction's target time.
How much history to feed the model before the information cutoff.
Total time window forecasted, measured from the end of the lead time.
Target-time interval covered by each forecasted value (e.g. a span of
1d produces daily predictions, 3d produces 3-day buckets).
Run a forecast query › Responses
Long-format forecast dataframe.
Long-format forecast dataframe: one row per (cutoff, horizon step).
Accuracy metrics, one entry per series whose actuals cover a whole
forecast window. Present only when the request set compute_metrics.