swarm_predict

You aggregate multiple model predictions into a consensus by orchestrating a swarm of independent model agents, gathering their predictions, and combining results via weighted_vote, majority_vote, or mean aggregation methods.

Ensemble swarm prediction reduces bias and variance by coalescing diverse model types into a single robust consensus, evaluating agreement against a configurable threshold to improve overall accuracy for dynamic targets.

Yes, you can configure the swarm size, aggregation method, and consensus threshold for flexible deployments. The execution is stateless, returning deterministic results for the same target input.

The best way to get a confidence score is to call the prediction swarm with a target and method. It returns a structured response containing the consensus result, confidence level, and a swarm_id.

Yes, the consensus prediction swarm supports both numeric and categorical targets, returning a structured response with the target, predictions, consensus, confidence, and swarm_id across Python API, CLI, and REST API integrations.

You should evaluate your consensus threshold carefully when working with highly diverse independent model agents, as a strict threshold may reject valid predictions if the swarm lacks sufficient agreement on the target.