Compute Weighted Residuals

Description

Computed weighted residuals from a model fit.

Usage

weighted.residuals(obj, drop0 = TRUE)

Arguments

Details

Weighted residuals are based on the working residuals and working weights, which for a lm fit are the raw residuals R_i multiplied by \sqrt{w_i}, where w_i are the weights as specified in lm's call.

For other models, including glm fits, weights and residuals are taken from the last iteration of the approximating weighted linear model fit. Notice that it is assumed that residuals() and weights() methods for obj accept type="working".

Dropping cases with weights zero is compatible with influence and related functions.

Value

Numeric vector of length n', where n' is the number of non-0 weights (drop0 = TRUE) or the number of observations, otherwise.

Note

Until R version 4.5, this returned deviance residuals, which are identical for lm fits, but differ for other models and are not well suited for influence measures.

See Also

residuals, lm.influence, etc.

Examples

## following on from example(lm)

all.equal(weighted.residuals(lm.D9),
          residuals(lm.D9))
x <- 1:10
w <- 0:9
y <- rnorm(x)
weighted.residuals(lmxy <- lm(y ~ x, weights = w))
weighted.residuals(lmxy, drop0 = FALSE)