| adding up time series question [message #45193] |
Mon, 15 August 2005 07:26 |
rlayberry
Messages: 33
Registered: November 2004
|
Member |
|
|
Hi
I have an interesting problem that I want to solve on idl which I want
to get some advice about....
I have 43 time series (thousands of points each). I want to add the
times series together with weigthing such that the standard deviation
of the summed time series is minimised.
For 2 time series
sd(a+b)^2=sd(a)^2+sd(b)^2 + 2*corr(a,b)*sd(a)*sd(b)
ie the standard deviation of 2 time series added together can be found
from the standard deviation of the individual time series and the
correlation between them.
if i add 43 time series together, is there a quick way of finding the
standard deviation of the summed time series from the sd and
correlation information from the individual time series?
basically, adding them up with different weights and finding the sd is
very time consuming. can i find the sd from the sd and corr data, or
is it just not possible to take a short-cut (i am tempted to think
this?)
thanks for any advice
russ
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
