Agreement between two models on the same quantity

Agreement between two models on the same quantity

Usage

agreement(reference, candidate, na.rm = TRUE)

Arguments

reference

Numeric vector from the reference model (the IBM).

candidate

Numeric vector from the model under test (fleet), the same length as reference.

na.rm

If TRUE, drop pairs in which either value is not finite.

Value

A one-row data frame with columns n, cor, rmse, bias, rel_bias and slope. bias is the mean signed difference, in the units of the quantity; rel_bias is bias divided by the mean of reference, so 0.087 means the candidate runs 8.7% high. slope is the OLS slope of candidate on reference, which can disagree with rel_bias: a slope of 1 with a positive bias is a constant offset, and a slope above 1 with zero bias is a fan. With fewer than two pairs every statistic is NA, and cor is NA when either input has no variance.