procrustes
,
promax
, the
oblimin
family, and the
orthomax
family.
factanal
,
princomp
.
rotate(x, ...) rotate.default(x, rotation=NULL, orthogonal=T, parameters=NULL, normalize=T)
rotate
, and a matrix (generally with
fewer columns than rows) for
rotate.default
.
"varimax"
,
"quartimax"
,
"equamax"
,
"parsimax"
,
"orthomax"
,
"covarimin"
,
"biquartimin"
,
"quartimin"
,
"oblimin"
,
"crawford.ferguson"
,
"procrustes"
,
"promax"
,
"none"
.
If
rotation
is
NULL
, then the remaining arguments define the rotation.
TRUE
, the rotation is constrained to be orthogonal.
"oblimin"
,
"orthomax"
and
"crawford.ferguson"
) will use them. This is required for Procrustes
rotations and must be a matrix of the same dimension as
x
.
If
parameters
is
NULL
, then the default for orthogonal rotations is
parameters=1
, and for oblique rotations the default is
parameters=c(1/p, 1, -1/p, -1)
where
p
is the number of rows of
x
.
TRUE
, then Kaiser normalization is performed.
rotate
that is used.
rotate.default
returns a list which depends on the type of rotation
(see
obliquemin
,
orthomax
and
procrustes
). The list that is returned
will always contain the following components:
rmat
up to numerical precision.
The
parameters
argument provides a means of selecting a specific rotation.
For orthogonal rotations (
rotation=orthomax
, or
rotation=NULL
and
orthogonal=T
),
parameters
must be of length 1.
For the oblimin family,
parameters
should be of length 1 or length 4.
Crawford-Ferguson rotations allows
parameters
to be of length 2 or 4.
Harman, H. H. (1976).
Modern Factor Analysis,
3rd Edition.
University of Chicago Press, Chicago.
Mardia, K. V., Kent, J. T. and Bibby, J. M. (1979).
Multivariate Analysis.
Academic Press, London.
amat <- matrix(c(2,-1,-2,3,2,1,-2,-4),4) rotate(amat) # varimax rotation rotate(amat, "quartimin") rotate(amat, ortho=T, param=10)