For every ELF and for ignition and escape, fireSenseUtils::storedFitProblem(): the fit's boosters must be Poisson counts and use only features among the covariates to predict (scaling does not change their names).

stopIfStoredFitsProblem(fits, covariates)

Arguments

fits

The result of ignitionFitsByELF().

covariates

The covariates to predict, a data.table.

Value

NULL, invisibly; stops with the ELF, process and reason of the first fit that has a problem.