Creating ARIMA model with more than one time series

Hi Everyone,

Thank you for taking the time to read my problem.

I have three time series: Ozone, Humidity, Temperature

I want to create an ARIMA model that will take into account these three time series, but the econometric modeller is only allowing one at a time which is not what i am after.

I have attached a screenshot of what the data looks like.

Please any suggestions would be of great help.

Answer 1

[Chidvi Modala]

The modeler application in Econometric toolbox supports only univariate (i.e. single variables) modelling and analysis. I have brought this to the concerned people and it might be considered in any future release. However, for multivariate MATLAB does have some other models

The model which can be chosen is based on the type of the data you have (i.e. Stationary or Non-stationary)

1. If you have a stationary data you can use "Vector AutoRegression(VAR)" model.

- You can find the details of this model here: https://www.mathworks.com/help/econ/varm.html#bvojt3n

- This documentation page also gives an example of how to estimate and forecast this model for 2 variables: https://www.mathworks.com/help/econ/varm.html#bvn5dwm-1

2. If you have a non-stationary data, then you can use "vector error-correction (VEC)" model.

- You can find the details of this model here: https://www.mathworks.com/help/econ/vecm.html?searchHighlight=vecm&s_tid=doc_srchtitle#mw_bfb5830f-cd78-4a52-9120-149fd56c8baf

- You can find the examples for this model here: https://www.mathworks.com/help/econ/vecm.html?searchHighlight=vecm&s_tid=doc_srchtitle#mw_75f0e2db-ba30-4cc2-b01b-baff69100a7e

[Saad Khan]

Thank you for giving me alternatives, appreciate it.