gendata: Generate Data for Simulation

In alexvolkmann/multifammPaper: Convenience Functions for the Analysis of the multiFAMM Paper

Generate Data for Simulation

Description

This function creates a new data set based on already a fitted model of type multiFAMM.

Usage

gendata(I = 10, J = 10, nested = FALSE, num_dim = 2, lamB, lamC, lamE,
  normal = TRUE, sigmasq = list(0.05, 4), dim_indep = TRUE,
  simple_sig = TRUE, mu = function(t) {
     t + sin(t)
 }, N_B = 1,
  N_C = 1, N_E = 1, phiB_fun, phiC_fun, phiE_fun, minsize = 10,
  maxsize = 20, min_grid = 0, max_grid = 1, grid_eval = 100,
  min_visit = 3, max_visit = 3, use_RI = FALSE, covariate = TRUE,
  num_cov = 4, interaction = FALSE, which_inter = matrix(NA),
  model = NULL, center_scores = FALSE, decor_scores = FALSE,
  nested_center = FALSE, trajectory = FALSE, equal_grid = FALSE)

Arguments

I	Number of levels for first grouping variable (individuals).
J	Number of levels for second grouping variable (set to NA for random intercept design).
nested	TRUE: second random effect is nested. FALSE: if it is a crossed effect. Defaults to FALSE.
num_dim	Number of dimensions.
lamB	Eigenvalues for first grouping variable.
lamC	Eigenvalues for second grouping variable. Can be NULL.
lamE	Eigenvalues for curves-specific deviations.
normal	TRUE: FPC weights are drawn from a normal distribution. FALSE: they are drawn from a mixture of normals. Defaults to TRUE.
sigmasq	Error variances for each dimension supplied as a list.
dim_indep	TRUE: Use rmvnorm to ensure that error variances are independent across the dimensions. FALSE: Prolonge Jona's implementation using . Defaults to TRUE.
simple_sig	TRUE: Random normal numbers weighted with sigma. FALSE: mvnorm. Defaults to TRUE.
mu	Mean function. Can be NULL if the GAM model is supplied, then defaults to t + sin(t). If GAM model is supplied, then this argument is redundant.
N_B	Number of FPCs for first grouping variable. Defaults to 1.
N_C	Number of FPCs for second grouping variable. Can be 0 and defaults to 1.
N_E	Number of FPCs for curves-specific deviations. Defaults to 1.
phiB_fun	Eigenfunctions of the first grouping variable as a multiFunData object.
phiC_fun	Eigenfunctions of the second grouping variable as a multiFunData object. Can be NULL.
phiE_fun	Eigenfunctions of the curve-specific deviations as a multiFunData object.
minsize	Minimal number of scalar observations per curve. Defaults to 10.
maxsize	Maximal number of observations per curve. Defaults to 20.
min_grid	Minimal value of grid range. Defaults to 0.
max_grid	Maximal value of grid range. Defaults to 1.
grid_eval	Length of the final grid to evaluate the functions. Defaults to 100.
min_visit	Minimal number of repetitions of each subject-word/session (first- and second grouping variable) combination (in case of random intercept design: minimal number of repetitions for first grouping variable). Defaults to 3.
max_visit	Maximal number of repetitions of each subject-word/session (first- and second grouping variable) combination (in case of random intercept design: maximal number of repetitions for first grouping variable). Defaults to 3.
use_RI	TRUE: data with a random intercept structure are generated. FALSE: data with crossed/nested random intercepts structure are generated (Additional layer "words"/"sessions" included). Defaults to FALSE.
covariate	If covariates shall be generated. Defaults to TRUE.
num_cov	Number of covariates if covariate = TRUE. Defaults to 4.
interaction	TRUE if there are interactions between covariates (as in the phonetic sparseFLMM). Defaults to FALSE.
which_inter	Symmetric matrix specifying the interaction terms (as in sparseFLMM). Defaults to missing matrix.
model	GAM model from which to extract the covariate functions. Can be NULL if mu is specified.
center_scores	TRUE: FPC weights are centered. Defaults to FALSE.
decor_scores	TRUE: FPC weights are decorrelated. Defaults to FALSE.
nested_center	TRUE: FPC weights of the nested effect are centered around zero for each subject. Defaults to FALSE.
trajectory	TRUE: All dimensions are observed at the same time points as would be (ideally) the case for the snooker data. Defaults to FALSE.
equal_grid	TRUE: Evaluation points are identical over the dimensions as would be for movement trajectories. Defaults to FALSE.

alexvolkmann/multifammPaper documentation built on Sept. 9, 2024, 8:47 p.m.