tvem (version 1.4.1)

select_tvem: select_tvem: Select number of interior knots for an unpenalized TVEM.

Description

select_tvem: Select number of interior knots for an unpenalized TVEM.

Usage

select_tvem(
  max_knots = 5,
  keep_going_if_too_few = FALSE,
  use_bic = FALSE,
  penalize = FALSE,
  print_output = TRUE,
  ...
)

Value

A TVEM object for the fitted model, with an additional component containing a table of information criteria.

Arguments

max_knots

The maximum number of interior knots to try (0 through max_knots)

keep_going_if_too_few

Whether to continue in a stepwise fashion if the max_knots does not seem to be high enough

use_bic

Whether to use BIC (TRUE) instead of AIC (FALSE) when selecting the best model. Note that both of these IC's are calculated from the working-independence pseudolikelihood rather than the unknown true likelihood. However, for BIC, the sample size is taken to be the number of subjects, not the number of observations.

penalize

Whether to include a penalty function in estimation

print_output

Whether to print the pseudolikelihoods obtained for each candidate number of interior knots.

...

Other inputs to be sent along to each call to the tvem function.

Examples

set.seed(123)
the_data <- simulate_tvem_example(n_subjects=200)
tvem_model <- tvem(data=the_data,
              formula=y~1,
              id=subject_id,
              time=time)
print(tvem_model)
plot(tvem_model)