Skip to main content
A process monitor may know the current load before it receives an alarm reading. Adapt-1 can forecast the final load and alarm together, then revise their probabilities when the alarm reading arrives. One measured outcome scores both forecasts. Use a UAT deployment that supports transition futures. Send the same Bearer token and session ID throughout the example. Paths are relative to the deployment’s API base, including /api/v1. For an executable Python client, see Run the examples.

Create the Domain

Send this body to POST /domains. Choose a fresh Domain ID and use it in the subsequent request paths.
Choose paths according to the actual payload, not the question text:
  • input_paths: information available when forecasting. Do not include the answer or anything computed from the eventual outcome.
  • targets: values measured later. Supported types are number, categorical, boolean and set. A target’s type describes its data, not its reward role.
  • event_types: which completed observations are eligible to train this learner.
  • group_by_paths: optional exact partitions, such as an instrument or process identity. Include the same grouping fields in queries and observations.
  • action_path: optional executed-action field. With it, outcomes are conditioned on that action, and planning can evaluate the declared alternatives.
Do not group by an episode ID or a unique request ID unless you really want each episode/request to have its own independent pool of examples. Support is required within the matching group and action, not just somewhere in the Domain. For time-based forecasting, define the target horizon explicitly in the data contract: for example, final_load means the load one hour after observation. If combining horizons in one learner, declare horizon/lead time as an input or grouping field and include it in training. The clock advancing alone is not new evidence. For automatic input selection, see Discover transition inputs.

Supply completed observations

Post historical observations to /domains/process-futures-example/events in time order. Send one request per measured transition. For example:
The example requires four matching observations. Until it has sufficient support, read the reason in transition_prediction.futures.abstention_reason.

Issue a forecast with partial evidence

POST /domains/process-futures-example/query:
Read transition_prediction.futures. The surrounding point forecast can abstain because an input is missing while the futures forecast remains available.

Revise with new evidence

Query again with the same future_id and all currently known inputs:
Every query returns a new decision_id. Keep future_id stable only for revisions of the same eventual outcome, with the same grouping and action. Send cumulative evidence: omitting an earlier input does not tell the server to remember it from the preceding request. Probabilities can concentrate, spread out, or reverse when evidence changes. Repeated identical queries do not manufacture confidence or new training samples.

Resolve the forecasts

POST /domains/process-futures-example/events:
Supply an actual returned decision ID from that future. This scores retained issue-time forecasts before using the newly observed target for learning. It resolves the outstanding revisions together; do not submit one duplicate outcome for every revision. Inspect learner_eligibility.futures_update.accepted. An eligible outcome without a decision link can still train transitions, but it cannot score a forecast that was never recorded. A successful HTTP response alone does not prove every learner accepted the row. For ordinary historical training, submit completed events in chronological order. An optional event timestamp is an ISO-8601 timestamp. For a historical online evaluation, reveal each outcome only after the corresponding simulated prediction and only when it would actually have become available. Backdating a leaked outcome does not undo that leakage. See Read the forecast distribution for the returned fields.