Effects from Fitted Model
Returns (orthogonal) effects from a fitted model, usually a linear
model. This is a generic function, but currently only has a methods for
objects inheriting from classes
## S3 method for class 'lm': effects(object, set.sign = FALSE, \dots)
- an Robject; typically, the result of a model fitting function
- logical. If
TRUE, the sign of the effects corresponding to coefficients in the model will be set to agree with the signs of the corresponding coefficients, otherwise the sign is arbitrary.
- arguments passed to or from other methods.
For a linear model fitted by
the effects are the uncorrelated single-degree-of-freedom values
obtained by projecting the data onto the successive orthogonal
subspaces generated by the QR decomposition during the fitting
process. The first $r$ (the rank of the model) are associated with
coefficients and the remainder span the space of residuals (but are
not associated with particular residuals).
Empty models do not have effects.
- A (named) numeric vector of the same length as
residuals, or a matrix if there were multiple responses in the fitted model, in either case of class
The first $r$ rows are labelled by the corresponding coefficients, and the remaining rows are unlabelled. Note that in rank-deficient models the corresponding coefficients will be in a different order if pivoting occurred.
Chambers, J. M. and Hastie, T. J. (1992) Statistical Models in S. Wadsworth & Brooks/Cole.
y <- c(1:3, 7, 5) x <- c(1:3, 6:7) ( ee <- effects(lm(y ~ x)) ) c( round(ee - effects(lm(y+10 ~ I(x-3.8))), 3) ) # just the first is different