Automatic ARIMA model identification in MATLAB (like auto.arima in R)

Is there a function in MATLAB to identify a suitable ARIMA model like it does in R (function of auto.arima in package forecast) or in other commercial forecasting products?

Answer 1

[Roger Wohlwend]

No, there is no such function. You can identify a suitable model with the Information criterions (AIC and BIC).

[Jonas]

There is some useful functions in System Identification Toolbox like " arx", that can find polynomial degree "p" of AR model.

• Can it help me to identify AR-part of ARMA/ARIMA model?
• And maybe SIT (Toolbox) contain other interesting function that can help to identify "p" and "q" for ARMA model?

[Sinan Islam]

Doesnt estimate do the same purpose as auto arima in R?

https://www.mathworks.com/help/econ/arima.estimate.html

[Nicholas]

No, an auto ARIMA function would not need any pre-specification of the p,d, q parameters. It would function like TimeSeriesModelFit https://reference.wolfram.com/language/ref/TimeSeriesModelFit.html in Wolfram for example.

Interestingly, when asked, CoPilot invents such a function (autoARIMA) but when one looks in the Econometrics toolbox it sadly doesn't exist. The code (that is apparently being hallucinated) is

% Load example data (e.g., airline passenger data)

load Data_Airline

y = log(Data); % Log-transform the data for stabilization

% Automatically fit an ARIMA model

Mdl = autoARIMA(y);

% Display the estimated model

disp(Mdl);