Random state space sampling predictor

Note

Check high level predictor docs for predictor basics.

This simple predictor, named Sampling, performs a random sampling of the state space, respecting the state space restrictions in Beam and direct dependencies of an application stack. It’s mostly suitable when dealing with cold start problem.

This predictor does not perform well for dependency graphs that are deep as it can spend a lot of time in state expansion. See random walk predictor as an alternative to this predictor.