simlong: Simulate longitudinal data on an equally spaced grid.
In wfforrest/maeve: Modeling Accessories Enabling Visualization of Experiments

Description Usage Arguments Value Author(s) References Examples

Allows simulation of a user-provided function on an equally spaced grid. Options for multiple groups and multiple subjects per group. Options for random dropout of observations, either binomially, geometric dropout, or a process based on an endpoint-driven hazard function so that once an observation is missed, that subject has no more observations. Also allows for linear random effects (random intecept and slope).

simlong(
  grid_length = 11,
  number_groups = 3,
  subjects_per_group = 5,
  SD_intercept = 1e-04,
  SD_slope = 1e-04,
  SD_epsilon = 1e-04,
  xmin = 0,
  xmax = 1,
  dropout_scheme = c("binomial", "geometric", "endpoint_dependent"),
  binom_retention_prob = 1,
  geom_dropout_prob = NULL,
  size_based_dropout = TRUE,
  dropout_map = NULL,
  force_dropout = TRUE,
  with_dropout_annot = FALSE,
  response_func = function(x, ...) {     6 * x * (1 - x) }
)

`grid_length`	integer number of equally-spaced points at which each subject is evaluated at the function.
`number_groups`	integer number of groups.
`subjects_per_group`	integer number of subjects per group.
`SD_intercept`	numeric Gaussian SD for random intercept term
`SD_slope`	numeric Gaussian SD for random slope term
`SD_epsilon`	numeric Gaussian SD for random within-subject "noise" term attached to each measurement.
`xmin`	numeric starting point of grid at which to evaluate the function for each subject.
`xmax`	numeric ending point of grid at which to evaluate the function for each subject.
`dropout_scheme`	character string specifying which dropout scheme is to be used.
`binom_retention_prob`	numeric probability of retaining an observation. Set between 0 and 1 for simple Bernoulli random thinning.
`geom_dropout_prob`	numeric geometric dropout probability in (0,1). Probability for subject to drop out at each post-baseline step. Set to non-NULL to override binomial thinning.
`size_based_dropout`	logical whether to determine dropout probability from current endpoint value.
`dropout_map`	data.frame mapping endpoint values to dropout probability values that can be used to interpolate dropout probability.
`force_dropout`	logical whether to force a dropout when the endpoint exceeds the upper end of the dropout map phenotype range.
`with_dropout_annot`	logical whether to skip filtering the data.frame and return everything along with columns to perform dropout filtering manually.
`response_func`	function to evaluate for each grid point, for each subject

data.frame with simulated data set.

Bill Forrest <forrest@gene.com>

Bill Forrest forrest@gene.com

www.r-project.org

simlong_example_00  = simlong( number_groups = 1, subjects_per_group = 1,
                               response_func = function( x, ... ) { 1 + x } )

simlong_example_01  = simlong( grid_length = 25, number_groups = 2, subjects_per_group = 3,
                               SD_intercept = .1, SD_slope = .1, SD_epsilon = .01, 
                               response_func = function( x, group ) { 
                                                   1 + x * scale( as.numeric( group ) )
                                                                    }   
                              )