Extracts Fisher information matrix of \(\Theta\) from objects returned by PLN and its variants. Fisher matrix is computed using one of two approximation scheme: wald (default, conservative, gives large confidence interval) or louis (anticonservative). Note that the Fisher information matrix is the full-data version (scaled by the number of observations), usually noted $$I_n(\theta)$$.
fisher(object, type)# S3 method for PLNfit
fisher(object, type = c("wald", "louis"))
A block-diagonal matrix with p (number of species) blocks of size d (number of covariates), assuming \(\Theta\) is a matrix of size d * p.
an R6 object with class PLNfit
Either wald (default) or louis. Approximation scheme used to compute the Fisher information matrix
fisher(PLNfit): Fisher information matrix for PLNfit
standard_error for standard errors