summary.edr {EDR} | R Documentation |
The function provides information on the estimated effective dimension reduction (EDR) space.
summary.edr(object, m = 1, R = NULL, ...)
object |
Object of class "edr" . |
m |
Dimension of the effective dimension reduction (EDR) space. m=1
corresponds to single index models, m>1 specifies a multiindex model.
Determines the number of eigenvectors and cumulative eigenvalues to show. |
R |
If code R specifies a matrix (dimension c(k,d) , k>=m,
d=dim(object$x)[2] , this matrix is interpreted as spanning the true
EDR space. Two distances between the estimated EDR space and the space spanned
R[1:m,] are computed. |
... |
Additional parameters will be ignored |
Provides information on the estimated effective dimension reduction (EDR) space.
The first m
basis vectors and the cummulative sum of normalized eigenvalues of matrix
object$bhat
are given. If R
is specified the distance
||R (I- hat{P}_m)||/||R|| mbox{with} hat{P}_m = U_m^T U_m , hat{R}_m=U_m Lambda V^T
and the distance specified by Li (1992) are computed.
Returns a list with components
Rhat |
(First) m eigenvectors of the estimated EDR space. |
cumlam |
Cummulative sum of first m eigenvalues of object$bhat , standardized by the sum of all eigenvalues. |
loss1 |
If R was specified the distance ||R (I- hat{P}_m)||/||R|| between the true and estimated
m-dimensional EDR space. |
loss2 |
The distance specified by Li (1992). |
Joerg Polzehl, polzehl@wias-berlin.de
edr
, edr.R
, print.edr
, plot.edr
require(EDR) ## Not run: demo(edr_ex1) ## Not run: demo(edr_ex2)