gamselBayes (version 2.0-1)

predict.gamselBayes: Obtain predictions from a gamselBayes() fit

Description

The estimated non-linear components of the generalized additive model selected via gamselBayes are plotted.

Usage

# S3 method for gamselBayes
predict(object,newdata,type = "response",...)

Value

A vector or data frame depending on the value of type:

If type="link" then the value is a vector of linear predictor-scale fitted values.

If type="response" and family="binomial" then the value is a vector of probability-scale fitted values. Otherwise (i.e. family="binomial") the value is the vector of predictor-scale fitted values.

If type="terms" then the value is a a data frame with number of columns equal to the total number of predictors. Each column is the contribution to the vector of linear predictor-scale fitted values from each predictor. These contributions do not include the intercept predicted value. The intercept predicted value is included as an attribute of the returned data frame.

Arguments

object

A gamselBayes() fit object.

newdata

A two-component list the following components:
A data frame containing new data on the predictors that are only permitted to have a linear or zero effect and, if not NULL, must have the same names as the Xlinear component of object.
A data frame containing new data for the predictors that are permitted to have a linear, nonlinear or zero effect and, if not NULL, must have the same names as the Xgeneral component of object.
If both Xlinear and Xgeneral are not NULL then they must have the same numbers of rows.

type

A character string for specifying the type of prediction, with the following options: "link", "response" or "terms", which leads to the value as described below.

...

A place-holder for other prediction parameters.

Author

Virginia X. He and Matt P. Wand

Examples

library(gamselBayes) 

# Generate some simple regression-type data:

n <- 1000 ; x1 <- rbinom(n,1,0.5) ; x2 <- runif(n) ; x3 <- runif(n) ; x4 <- runif(n)
y <- x1 + sin(2*pi*x2) - x3 + rnorm(n)
Xlinear <- data.frame(x1) ; Xgeneral <- data.frame(x2,x3,x4)
names(Xlinear) <- c("x1") ; names(Xgeneral) <- c("x2","x3","x4")

# Obtain and summarise a gamselBayes() fit for the data:

fit <- gamselBayes(y,Xlinear,Xgeneral)
summary(fit)   

# Obtain some new data:

nNew <- 10
x1new <- rbinom(nNew,1,0.5) ; x2new <- runif(nNew) ; x3new <- runif(nNew) 
x4new <- runif(nNew)
XlinearNew <- data.frame(x1new) ; names(XlinearNew) <- "x1"
XgeneralNew <- data.frame(x2new,x3new,x4new)
names(XgeneralNew) <- c("x2","x3","x4")

newdataList <- list(XlinearNew,XgeneralNew)

# Obtain predictions at the new data:

predObjDefault <- predict(fit,newdata=newdataList)
print(predObjDefault)