Two-fold cross-validation of the spread fit

Fits the model to every other year and simulates the held-out years from the simulateMembers best members of that fit, without the size cap, then the same the other way round. Uses the same objective settings as the run event. Writes nothing to the ledger.

crossValidateSpread(sim, covs, thresh)

Arguments

sim

a simList.

covs

mod$covsX1000.

thresh

mod$thresh.

Value

list: sims (from fireSenseUtils::simulateFireSizes(), with fold) and score (from fireSenseUtils::scoreFireSizes() on both folds together, so every year is predicted once).