| ** SVD using WAVE-Advantage ** [message #2091] |
Tue, 17 May 1994 07:52 |
jhai
Messages: 1
Registered: May 1994
|
Junior Member |
|
|
Hello:
I am doing a SVD of a rectangular array and having problems
understanding the output. (I am no expert on SVD).
WAVE> a=fltarr(581,499)
WAVE> info,a
A FLOAT = Array(581, 499)
WAVE> s=svdcomp(a,U=u,V=v)
WAVE> info,a,u,v,s
A FLOAT = Array(581, 499)
U FLOAT = Array(581, 499)
V FLOAT = Array(499, 499)
S FLOAT = Array(499)
WAVE>
According to my understanding of the SVD, the dimensions of U should be
(499, 499) and that of V should be (581, 499)
Actually I am using the SVD to get the principal componenets and pc-scores.
After I have got the U and V matrices, the PC-scores can be obtained by
multiplying the columns of the U matrix by the corresponding singular
value. In the above example I have 499 sv's and U has 581 columns!!
Can somebody correct me where I am going wrong.
Thanks.
the columns of U multiplied by the
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
