predict.cozigam {COZIGAM}R Documentation

Prediction from a fitted COZIGAM

Description

Predict from a COZIGAM fitted by the cozigam function, given a new set of covariate values, if supplied, or the original set of covariate values used for the model fit. The standard errors of the point predictors are also computed.

Usage

## S3 method for class 'cozigam':
predict(object, newdata, type="link", se.fit=FALSE, ...)

Arguments

object A fitted cozigam object as produced by the cozigam function.
newdata A data frame containing the values of the model covariates at which predictions are required. If this is not provided then predictions corresponding to the original data are returned. If newdata is provided then it must contain all the variables needed for prediction.
type When this has the value "link" (default) the linear predictor (possibly with associated standard errors) is returned. When type="terms" each component of the linear predictor is returned seperately (possibly with standard errors): this excludes any intercept. When type="response" predictions on the scale of the response are returned (possibly with approximate standard errors).
se.fit Logical. If TRUE (not default), the standard errors are returned.
... Other arguments.

Details

The standard errors produced by predict.cozigam() are based on the covariance matrix of the parameters obtained by Louis' method in the fitted gam object.

Value

If se.fit is TRUE then a three-item list is returned with items (both arrays) fit, se.fit containing the point predictors and associated standard error estimates and p containing the corresponding zero-inflation rates, otherwise a two-item list without the array of standard error estimates is returned. The dimensions of the returned arrays depends on whether type is "terms" or not: if it is then the array is 2-dimensional with each term in the linear predictor included separately, otherwise the array is 1-dimensional and contains the linear predictor/predicted values (or corresponding standard errors). The linear predictor returned termwise excludes the intercept.

Author(s)

Hai Liu and Kung-Sik Chan

References

Liu, H and Chan, K.S. (2008) Constrained Generalized Additive Model with Zero-Inflated Data. Technical Report, Department of Statisics and Actuarial Science, The Unversity of Iowa. http://www.stat.uiowa.edu/techrep/tr388.pdf

Louis, T. A. (1982) Finding the Observed Information Matrix When Using EM Algorithm. J. R. Statist. Soc. B, 44, 226-233

See Also

cozigam,plot.cozigam

Examples

set.seed(11)
n <- 600
x1 <- runif(n, 0, 1)
x2 <- runif(n, 0, 1)
x3 <- runif(n, 0, 1)

f <- testfn(x1, x2)*4-mean(testfn(x1, x2)*4) + f0(x3)/2-mean(f0(x3)/2)
sig <- 0.5
mu0 <- f + 3
y <- mu0 + rnorm(n, 0, sig)

alpha0 <- -2.2
delta0 <- 1.2
p0 <- .Call("logit_linkinv", alpha0 + delta0 * mu0, PACKAGE = "stats")
z <- rbinom(rep(1,n), 1, p0)
y[z==0] <- 0

res <- cozigam(y~s(x1,x2)+s(x3), conv.crit.out = 1e-4, family = gaussian)

newdata <- data.frame(x1=c(0.5,0.8), x2=c(0.2,0.1), x3=c(0.3,0.7))
pred <- predict(res, newdata=newdata, se.fit=TRUE, type="response")

[Package COZIGAM version 1.0-1 Index]