Legacy Documentation

Time Series (2011)

This is documentation for an obsolete product.
Current products and services

User's Guide to Time Series / Forecasting /
Previous section-----Next section

1.7.6 Forecasting for Multivariate Time Series

The above forecast formulas and functions for univariate time series can be generalized directly to the multivariate case. Here the forecast is a vector, and the mean square errors are now error covariance matrices and (7.8) becomes . The function newforecast we wrote earlier for updating the forecast should be modified to incorporate the matrix nature of the and weights.
This defines newforecast for the multivariate case.
In[44]:=
Example 7.10 Forecast a bivariate ARMA(1, 1) model with forecast origin 30 and lead time 3.
This defines the model.
In[45]:=
The random number generator is seeded.
In[46]:=
This gives the prediction for the next three values of the series and the error covariance matrices.
In[47]:=
Out[47]=