General Nonparametric Jackknife for Fitted Models

DESCRIPTION:

Performs delete-one jackknifing of observations from fitted linear models of class lm, glm, censorReg, or censorRegList. Calculates jackknife statistics for parameters of interest and produces an object of class .

USAGE:

jackknife(data, statistic, ...) 
jackknife.lm(data, statistic, ...) 
jackknife.glm(data, statistic, ...) 
jackknife.censorReg(data, statistic, ...) 

REQUIRED ARGUMENTS:

data
object to be jackknifed, of class , , , or censorRegList.
statistic
statistic to be jackknifed; a function or expression that operates on lm, glm, or censorReg objects and returns a vector or matrix. It may be a function (e.g. coef) which accepts data as the first argument; other arguments may be passed using args.stat.
Or it may be an expression such as predict(fit,newdata=orig.frame). If the data object is given by name (e.g. data=fit) then use that name in the expression, otherwise (e.g. data=glm(formula,dataframe)) use the name data in the expression, e.g. predict(data,newdata=orig.frame).

OPTIONAL ARGUMENTS:

The remaining arguments have the same usage as in the default version of .

VALUE:

an object of class which inherits from resamp. See for more details.

BUGS:

See .

SEE ALSO:

, , , , , ,

EXAMPLES:

# jackknife and lm 
fit <- lm(Fuel~Weight+Disp., data=fuel.frame) 
jackknife(fit, coef) 
 
# jackknife and glm 
fit <- glm(ozone^(1/3) ~ bs(radiation, 5,knots=185,Boundary.knots=c(7,334)) + 
           (wind + temperature)^2, data=air) 
newdata <- air 
jackknife(fit, predict(fit, newdata=newdata, type="link")) 
 
# jackknife and censorReg (use default statistic) 
fit <- censorReg(censor(days, event) ~ voltage, data = capacitor2, 
weights=weights, distribution = "weibull", threshold = "Lin") 
jackknife(fit, coef(fit)) 
 
# See help(bootstrap.lm) for more examples of syntax.