try.vglm: Silently Catch Errors for 'vglm()' Calls

In jenniferthompson/ClusterBootMultinom: Multinomial Logistic Regression Using Clustered Bootstrapping

Description

This function is used in multi.bootstrap() to catch errors and warnings in vglm() model fits.

Usage

try.vglm(...)

Arguments

...	Arguments to be passed to vglm().

Value

If fit is successful, an object of class vglm. If fit is unsuccessful, an object of class try-error.

See Also

vglm(); try().

Examples

df <- data.frame(id = sample(1:20, size = 100, replace = TRUE),
                 x1 = rnorm(n = 100),
                 x2 = rbinom(p = 0.75, n = 100, size = 1),
                 y = sample(LETTERS[1:3], size = 100, replace = TRUE))
df <- df[order(df$id),]
df$time <- unlist(lapply(1:length(unique(df$id)),
                         FUN = function(idnum){ 1:nrow(df[df$id == unique(df$id)[idnum],]) }))

## Successful fit
try.vglm(y ~ x1 + x2, data = df, family = multinomial(refLevel = 1))

## Unsuccessful fit
try.vglm(y ~ x1 + x2, data = df[1:5,], family = multinomial(refLevel = 1))

jenniferthompson/ClusterBootMultinom documentation built on May 19, 2019, 4:03 a.m.