Skip to main content

Module inv_gauss_link

ndarray_glm

Module inv_gauss_link 

Source
Expand description

Link functions for inverse Gaussian regression

Structsยง

Log
The log link $g(\mu) = \log(\mu)$ avoids linear predictors that give negative expectations.
NegRecSq
The canonical link function for inverse Gaussian regression is $g_0(\mu) = -1/(2\mu^2)$. This fails to prevent negative predicted y-values.