EViews.com

Posted: **Tue Nov 16, 2010 7:22 am**

Could you point me the optimal way of calculating weighted average of time series without using loops?
There is a group (A) of indices available and another group (B) containing some absolute values i.e weights. The group of weights (B) should be normalized first. The number of time series to be weighted is not known a priori...

=(A.1*B.1 + A.2*B.2 + A.3*B.3 + ...)/sum(B)

Posted: **Tue Nov 16, 2010 7:57 am**

I don't think you can do it without loops.

Posted: **Tue Nov 16, 2010 8:47 am**

yes it can be done, I figured it out myself...

Code: Select all


series a1 = rnd
series a2 = rnd
series a3 =rnd

series b1 = rnd
series b2 = rnd
series b3 =rnd

group  a  a* 
group  b  b* 

%c = "("+ @wdelim(@replace(@winterleave(@wcross(A.@members, "* "),b.@members),"* ","*")," ","+") +")/@rsum(B)" 
series w_avg  = {%c}

Posted: **Tue Nov 16, 2010 9:06 am**

Ha, fair enough.

EViews.com

Weighted average of group of time series

Weighted average of group of time series

Weighted average of group of time series

Re: Weighted average of group of time series

Re: Weighted average of group of time series