Does MATLAB do ARIMA models?

I would like to know if MATLAB does autoregressive integrated moving average modeling.

Answer 1

[MathWorks Support Team]

This enhancement has been incorporated in Release 2012a (R2012a). For previous product releases, read below for any possible workarounds:

Neither MATLAB, nor any of the toolboxes, contain a function that does ARIMA modeling. However, there are two possible workarounds:

- the System Identification toolbox does Box-Jenkins modeling and ARIMA models are a type of Box-Jenkins models. Therefore, you may be able to use the System Identification Toolbox to create your own ARIMA model.

Reference for ARIMA models are:

Quantitative Forecasting Methods by Nicholas R. Farnum and LaVerne W. Stanton

Neural Networks in Finance and Investing by Robert. R. Trippi and Efraim Turbon

Time Series Analysis by James D. Hamilton

- ARIMA model can be implemented using the GARCH functionality in the Econometrics Toolbox. For example, in order to implement ARIMA(1,1,1) model create an ARIMA111.m file containing the following code:

function pred = ARIMA111(y)

    predictions=1;

    p=1;

    q=1;

    yy = y(1);

    y = diff(y);

    Spec = garchset('R',p,'M',q,'C',NaN,'VarianceModel','Constant','Display','Off'); 

    [EstSpec,EstSE] = garchfit(Spec,y); 

    [sigmaForecast,meanForecast] = garchpred(EstSpec,y,predictions);

    pred = cumsum([yy; y; meanForecast])

    pred = pred(end-predictions+1:end);

end

You can then obtain one prediction from the model in the return value:

y=[6947.5;    5867.9;    4581.4;    7051.6;    6895.6;    6925.3];

ARIMA111(y)