gamsim {vagam}

Simulate example datasets from a generalized additive models (GAM).

Description

This function is a modification from example 7 of the gamSim function available in the mgcv package (Wood, 2017), which is turn is Gu and Wahba 4 univariate example with correlated predictors. Please see the source code for exactly what is simulated. The function is primarily used as the basis for conducting the simulation studies in Hui et al., (2018).

Usage

gamsim(n = 400, extra.X = NULL, beta = NULL, dist = "normal", scale = 1, offset = NULL)

Arguments

n

Sample size.

extra.X

Extra covariates, including critically an intercept if is to be included in the linear predictor for the GAM.

beta

Regression coefficient estimates.

dist

Currently only the "normal", "poisson" or "binomial" corresponding to the binomial distributions are available.

scale

Scale parameter in the Normal distribution.

offset

This can be used to specify an a-priori known component to be included in the linear predictor during fitting. This should be NULL or a numeric vector of length equal to n.

Value

A data frame containing information such as the simulated responses, covariates, each of the 4 "truth" smooths, and the overall linear predictor.

Author(s)

NA

References

• Hui, F. K. C., You, C., Shang, H. L., and Mueller, S. (2018). Semiparametric regression using variational approximations, Journal of the American Statistical Association, forthcoming.

• Wood, S. N. (2017) Generalized Additive Models: An Introduction with R (2nd edition). Chapman and Hall/CRC.

See Also

vagam for the main fitting function

Examples

normal_dat = gamsim(n = 40, dist = "normal", 
                    extra.X = data.frame(int = rep(1,40), trt = rep(c(0,1), each = 20)),
                    beta = c(-1, 0.5))

pois_dat = gamsim(n = 40, dist = "poisson", 
                  extra.X = data.frame(int = rep(1, 40), trt = rep(c(0,1), each = 20)),
                  beta = c(-1, 0.5))

binom_dat = gamsim(n = 40, dist = "binomial",
                   extra.X = data.frame(int = rep(1, 40), trt = rep(c(0,1), each = 20)),
                   beta = c(0, 0.5))

## Please see examples in the help file for the vagam function.