Mean chi^2 vs chi^2

[dimalvovs]

Why meanChisq is so different to the iteration chisq? On the tiny dataset it's about 20% difference, but we saw much larger differences.

To reproduce:

data(GIST)
data <- GIST.data_frame
unc <- 0.1*as.matrix(data)
res <- CoGAPS(data, nIterations=1000, uncertainty=unc, nPatterns=3,
                seed=1, messages=FALSE, sparseOptimization=FALSE)
res@metadata$chisq
res@metadata$meanChiSq

yields

>     res@metadata$chisq
[1] 9639.871 9359.936
>     res@metadata$meanChiSq
[1] 7466.165