ARMA model order
Purpose
Description
Tips
Macro Synopsis
Modules
Related Functions
References
Purpose
Estimates the optimum ARMA model order
Description
"ARMA model order" computes the optimum order of an
Auto-Regressive Moving Average process to model the given signal. This
is done by means of minimizing one of the following information criteria:
- Akaike's Information Criterion,
- the Bayesian Information Criterion,
- the Hannan-Quinn Criterion,
N is the signal length in samples and 
denotes
the variance of the residuals of the ARMA[p,q] process.
Parameters of "ARMA model order" are
- method, which is one of
- "Akaike" (0),
- "Bayesian" (1) or
- "Hannan-Quinn" (2)
- maximum AR order Pmax and
- maximum MA order Qmax.
The resulting model orders and the ARMA coefficients a and b of the
optimal order are returned in the message window. The resulting signal
shows the information criterion for the optimal MA order q displayed
versus the AR order p.
Tips
The Bayesian and Hannan-Quinn methods tend to give better results
and smaller model orders than Akaike's method.
Macro Synopsis
y = ARMAorder(x,method,Pmax,Qmax);
signal x,y;
option method;
int Pmax, Qmax;
Modules
Statistics
Related Functions
AR simulation, ARIMA modeling,
ARIMAX modeling, ARMA modeling,
ARMA simulation, ARMA spectral density,
ARMAX modeling,
ARMAX simulation, NAR modeling,
NARMA modeling.
References
Conover [11], Graybill [12],
Hollander/Wolfe [13], Schlittgen/Streitberg [14].