generate_data: Data generation function

In CIEE: Estimating and Testing Direct Effects in Directed Acyclic Graphs using Estimating Equations

Description

Function to generate data with n observations of a primary outcome Y, secondary outcome K, exposure X, and measured as well as unmeasured confounders L and U, where the primary outcome is a quantitative normally-distributed variable (setting = "GLM") or censored time-to-event outcome under an accelerated failure time (AFT) model (setting = "AFT"). Under the AFT setting, the observed time-to-event variable T=exp(Y) as well as the censoring indicator C are also computed. X is generated as a genetic exposure variable in the form of a single nucleotide variant (SNV) in 0-1-2 additive coding with minor allele frequency maf. X can be generated independently of U (X_orth_U = TRUE) or dependent on U (X_orth_U = FALSE). For more details regarding the underlying model, see the vignette.

Usage

generate_data(setting = "GLM", n = 1000, maf = 0.2, cens = 0.3,
  a = NULL, b = NULL, aXK = 0.2, aXY = 0.1, aXL = 0, aKY = 0.3,
  aLK = 0, aLY = 0, aUY = 0, aUL = 0, mu_X = NULL, sd_X = NULL,
  X_orth_U = TRUE, mu_U = 0, sd_U = 1, mu_K = 0, sd_K = 1, mu_L = 0,
  sd_L = 1, mu_Y = 0, sd_Y = 1)

Arguments

setting	String with value "GLM" or "AFT" indicating whether the primary outcome is generated as a normally-distributed quantitative outcome ("GLM") or censored time-to-event outcome ("AFT").
n	Numeric. Sample size.
maf	Numeric. Minor allele frequency of the genetic exposure variable.
cens	Numeric. Desired percentage of censored individuals and has to be specified under the AFT setting. Note that the actual censoring rate is generated through specification of the parameters a and b, and cens is mostly used as a check whether the desired censoring rate is obtained through a and b (otherwise, a warning is issued).
a	Integer for generating the desired censoring rate under the AFT setting. Has to be specified under the AFT setting.
b	Integer for generating the desired censoring rate under the AFT setting. Has to be specified under the AFT setting.
aXK	Numeric. Size of the effect of X on K.
aXY	Numeric. Size of the effect of X on Y.
aXL	Numeric. Size of the effect of X on L.
aKY	Numeric. Size of the effect of K on Y.
aLK	Numeric. Size of the effect of L on K.
aLY	Numeric. Size of the effect of L on Y.
aUY	Numeric. Size of the effect of U on Y.
aUL	Numeric. Size of the effect of U on L.
mu_X	Numeric. Expected value of X.
sd_X	Numeric. Standard deviation of X.
X_orth_U	Logical. Indicator whether X should be generated independently of U (X_orth_U = TRUE) or dependent on U (X_orth_U = FALSE).
mu_U	Numeric. Expected value of U.
sd_U	Numeric. Standard deviation of U.
mu_K	Numeric. Expected value of K.
sd_K	Numeric. Standard deviation of K.
mu_L	Numeric. Expected value of L.
sd_L	Numeric. Standard deviation of L.
mu_Y	Numeric. Expected value of Y.
sd_Y	Numeric. Standard deviation of Y.

Value

A dataframe containing n observations of the variables Y, K, X, L, U. Under the AFT setting, T=exp(Y) and the censoring indicator C (0 = censored, 1 = uncensored) are also computed.

Examples

# Generate data under the GLM setting with default values
dat_GLM <- generate_data()
head(dat_GLM)

# Generate data under the AFT setting with default values
dat_AFT <- generate_data(setting = "AFT", a = 0.2, b = 4.75)
head(dat_AFT)

CIEE documentation built on May 2, 2019, 6:39 a.m.