Transform the linear predictor to predicted values/probabilities.
Examples
x <- rnorm(n = 10L)
.mean_function(x, family = "binomial")
#> [1] 0.3987606 0.5251807 0.7151961 0.8114708 0.4020712 0.3815242 0.4215974
#> [8] 0.7702996 0.3201722 0.3431547
.mean_function(x, family = "poisson")
#> [1] 0.6632311 1.1060646 2.5111874 4.3042169 0.6724399 0.6168781 0.7288997
#> [8] 3.3534979 0.4709607 0.5224284