deviances {RCM} | R Documentation |
A function to extract deviances for all dimension, including after filtering on confounders
deviances(rcm, squaredSum = FALSE)
rcm |
an object of the RCM class |
squaredSum |
a boolean, should total deviance be returned? Total deviances can be deceptive and not correspond to the differences in log-likelihood. As the dispersion is different for each model. To compare models it is better to compare likelihoods. |
If Sum is FALSE, a named array of deviance residuals of the independence model and all models with dimension 1 to k, including after filtering on confounders. Otherwise a table with total deviances (the sum of squared deviance residuals), deviance explained and cumulative deviance explained.