glm Objects
glm objects.
residuals.glm(object, type)This is a method for the function
residuals() for objects inheriting
from class
"glm". However, as several types of residuals are
available for
glm objects, there is an additional optional argument
type.
"glm" representing a fitted model.
"deviance",
"pearson",
"working"
or
"response"; the first is the default.
y-fitted(object). The
summary() method for
glm objects produces deviance residuals.
The
residuals component of a
glm object contains the working residuals.
Chambers, J.M., and Hastie, T.J. (1991). Statistical Models in S, pp. 204--206.
kyph.glm <- glm(Kyphosis ~ ., binomial, kyphosis)
plot(fitted(kyph.glm), resid(kyph.glm, type = "response"), ylab =
"response residuals")
#produce a matrix of partial residuals, one column for each term
kyph.predict <- predict(kyph.glm, type = "terms") + kyph.glm$residuals
par(mfrow=c(1,3))
plot(kyphosis$Age, kyph.predict[, "Age"], ylab = "partial residuals")
plot(kyphosis$Start, kyph.predict[,"Start"], ylab="")
plot(kyphosis$Number, kyph.predict[,"Number"],ylab="")