[Gretl-users] Classical Decomposition

[Ignacio Diaz-Emparanza]

El Friday 01 February 2008 11:50:41 Tom La Bone escribió:
> Classical decomposition is breaking a univariate times series down into a
> trend component, a seasonal component, and a random noise component (for
> example, see Chapter 1 of Brockwell and Davis intro to time series book).
>
> Tom

You have the Hodrick-Prescott and Baxter-King filters to get the trend. We 
currently don't have a native decomposition procedure.

I wrote a gretl package for decomposing a time series based in the structural 
models framework (see Harvey's "Time series Models" or "Forecasting 
structural time series models and the Kalman filter"). It is in a very "beta" 
state, so it is not finished and it is very slow (I will send it to you in a 
private message). 

Jack is working in a user interface for the Kalman filter which may do this 
things to estimate faster.
But it is work in progress ...





-- 
Ignacio Diaz-Emparanza  
DEPARTAMENTO DE ECONOMÍA APLICADA III (ECONOMETRÍA Y ESTADÍSTICA)                                        
UPV/EHU
Avda. Lehendakari Aguirre, 83 | 48015 BILBAO
T.: +34 946013732 | F.: +34 946013754
www.et.bs.ehu.es