Generate an artificial longitudinal data set.
simulate_data(
N,
t_data,
covariates = c(),
names = NULL,
relevances = c(1, 1, rep(1, length(covariates))),
n_categs = rep(2, sum(covariates %in% c(2, 3))),
t_jitter = 0,
lengthscales = rep(12, 2 + sum(covariates %in% c(0, 1, 2))),
f_var = 1,
noise_type = "gaussian",
snr = 3,
phi = 1,
gamma = 0.2,
N_affected = round(N/2),
t_effect_range = "auto",
t_observed = "after_0",
c_hat = 0,
dis_fun = "gp_warp_vm",
bin_kernel = FALSE,
steepness = 0.5,
vm_params = c(0.025, 1),
continuous_info = list(mu = c(pi/8, pi, -0.5), lambda = c(pi/8, pi, 1)),
N_trials = 1,
force_zeromean = TRUE
)Number of individuals.
Measurement times (same for each individual, unless
t_jitter > 0 in which case they are perturbed).
Integer vector that defines the types of covariates (other than id and age). If not given, only the id and age covariates are created. Different integers correspond to the following covariate types:
0 = disease-related age
1 = other continuous covariate
2 = a categorical covariate that interacts with age
3 = a categorical covariate that acts as a group offset
4 = a categorical covariate that that acts as a group offset AND is restricted to have value 0 for controls and 1 for cases
Covariate names.
Relative relevance of each component. Must have be a vector
so that length(relevances) = 2 + length(covariates).
First two values define the relevance of the individual-specific age and
shared age component, respectively.
An integer vector defining the number of categories
for each categorical covariate, so that length(n_categs) equals to
the number of 2's and 3's in the covariates vector.
Standard deviation of the jitter added to the given measurement times.
A vector so that length(lengthscales) =
2 + sum(covariates %in% c(0,1,2)).
variance of f
Either "gaussian", "poisson", "nb" (negative binomial), "binomial", or "bb" (beta-binomial).
The desired signal-to-noise ratio. This argument is valid
only when noise_type is "gaussian".
The inverse overdispersion parameter for negative binomial data.
The variance is g + g^2/phi.
The dispersion parameter for beta-binomial data.
Number of diseased individuals that are affected by the
disease. This defaults to the number of diseased individuals. This argument
can only be given if covariates contains a zero.
Time interval from which the disease effect times are sampled uniformly. Alternatively, This can any function that returns the (possibly randomly generated) real disease effect time for one individual.
Determines how the disease effect time is observed. This
can be any function that takes the real disease effect time as an argument
and returns the (possibly randomly generated) observed onset/initiation time.
Alternatively, this can be a string of the form "after_n" or
"random_p" or "exact".
a constant added to f
A function or a string that defines the disease effect. If
this is a function, that function is used to generate the effect.
If dis_fun is "gp_vm" or "gp_ns", the disease component is drawn from
a nonstationary GP prior ("vm" is the variance masked version of it).
Should the binary kernel be used for categorical
covariates? If this is TRUE, the effect will exist only for group 1.
Steepness of the input warping function. This is only used if the disease component is in the model.
Parameters of the variance mask function. This is only
needed if useMaskedVarianceKernel = TRUE.
Info for generating continuous covariates. Must be a
list containing fields lambda and mu, which have length 3.
The continuous covariates are generated so that x <- sin(a*t + b) + c,
where
t <- seq(0, 2*pi, length.out = k)
a <- mu[1] + lambda[1]*stats::runif(1)
b <- mu[2] + lambda[2]*stats::runif(1)
c <- mu[3] + lambda[3]*stats::runif(1)
The number of trials parameter for binomial data.
Should each component (excluding the disease age component) be forced to have a zero mean?
An object of class lgpsim.