Displaying 1 result from an estimated 1 matches for "pi1fore".
2012 Aug 27
0
How can I find the principal components and run regression/forecasting using dynlm
...n earlier; h is the forecasting horizon; since dynlm
can't work for zoo object, the variables need to be transformed to ts
objects.
Xlast<-as.matrix(X[(t-h):(t-h-i+1)])
zlast<-t(as.matrix(z[(t-h):(t-h-j+1)])) # Make it a column matrix
beta<-as.matrix(model$coefficient)
pi1fore[t]<-pi1[t-h]+c(1, Xlast, zIlast)%*%beta # pi1fore is defined
earlier . I just use the inner product as a vehicle for forecasting. I
don't know if there's a way that the codes can be written neatly?
return(pi1fore)
Then I attempt to program similarly for the multi-factor case:...