Kalman filtering in [R] : FKF package and DLM

Asked Jul 27 '15 at 14:13

Active Jul 27 '15 at 19:26

Viewed 737 times

I am trying to implement a time varying state-space model in [R]. Model includes some exogenous variables that are part of the measurement and transition matrices.

I tried multiple packages and my final candidates are to FKF and DLM. However, I have a problem:

In DLM when I try to add seasonality to the model, the code becomes deadly slow. My dataset includes 10 years weekly observations and there is annual seasonality pattern in values. Therefore, I used + dlmModSeas(51) to introduce it. However, I guess it adds 51 dummy variables to the model that makes it tremendously slow. Question: How to increase the speed of DLM? Do I make a mistake in using seasonality or it is inherently slow?

I highly appreciate your kind help in advance,

Yours, Hamid

edited Jul 27 '15 at 19:26

asked Jul 27 '15 at 14:13

Hamid R. Darabi

1

Can you make your question explicit? (Eg, I don't see any question marks.) Are you just asking for help with speeding up R code? – gung - Reinstate Monica Jul 27 '15 at 14:37

Kalman filtering in [R] : FKF package and DLM

0 Answers0