How can I provide all the elements of a list to a function that needs seperate objects?-CodePudding

I'd like to be able to use the LRstats function from the vcdExtra package to compare 50 models. However, this function appears to be set up such that it needs each model object provided individually. If I have a list of 50 models, how can I do this without having to specify each model manually i.e. test[[1]], test[[2]], test[[3]]...etc?

library(vcdExtra)
library(dplyr)


test <- data.frame(
  y = rbinom(2500, 1, 0.5),
  x1 = rnorm(2500, mean = 0, sd = 1),
  x2 = rnorm(2500, mean = 1, sd = 3),
  z = rep(seq(from = 400, length.out = 50, by = 400), times = 50))


test_list <- group_split(test, z)
names(test_list) <- unique(test$z)



test_models <- lapply(test_list, function(x) glm(y ~ x1   x2, data = x, family = "binomial"))


test_models2 <- glmlist(test_models)

>Warning message:
In glmlist(test_models) :
  Objects test_models removed because they are not glm objects

test_LR <- LRstats(test_models2)
>NULL

CodePudding user response：

The problem was in the way function glmlist was used. glmlist expects one or more objects inheriting from class "glm" or "loglm". From the documentation:

Any objects which do not inherit the appropriate class glm or loglm are excluded, with a warning.

So call the function on each list member with do.call.

library(vcdExtra)
#> Loading required package: vcd
#> Loading required package: grid
#> Loading required package: gnm

test <- data.frame(
  y = rbinom(2500, 1, 0.5),
  x1 = rnorm(2500, mean = 0, sd = 1),
  x2 = rnorm(2500, mean = 1, sd = 3),
  z = rep(seq(from = 400, length.out = 50, by = 400), times = 50))

test_list <- dplyr::group_split(test, z)
names(test_list) <- unique(test$z)

test_models <- lapply(test_list, function(x) glm(y ~ x1   x2, data = x, family = "binomial"))

test_models2 <- do.call(glmlist, test_models)
test_LR <- LRstats(test_models2)
head(test_LR)
#> Likelihood summary table:
#>         AIC    BIC LR Chisq Df Pr(>Chisq)  
#> 400  72.136 77.872   66.136 47    0.03420 *
#> 800  68.608 74.344   62.608 47    0.06340 .
#> 1200 74.842 80.578   68.842 47    0.02056 *
#> 1600 73.545 79.281   67.545 47    0.02634 *
#> 2000 73.514 79.250   67.514 47    0.02649 *
#> 2400 70.524 76.260   64.524 47    0.04564 *
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

^{Created on 2022-03-31 by the reprex package (v2.0.1)}