Analysis configuration

Select the following option from menu to run the VAR model:

Analyze > Forcasting > Mutivariate Time Series > Vector Autoregressive Models

For the analysis, the following configurations are done in the user interface.
Variables tab
Endogenous Variables Exogenous Variables
  • Sales of Men's Clothing [men]
  • Sales of Women’s Clothing [women]
  • Sales of Jewelry [jewel]
  • Number of Catalogs Mailed [mail]
  • Number of Pages in Catalog [page]
  • Number of Phone Lines Open for Ordering [phone]
For the Lag Selection, select AIC, BIC, and HQIC. Enter the Maximum lag as 2.
Statistics tab
Select the following options under Fit Measures and Forecasts.
  • Descriptive Statistics
  • Goodness of Fit
  • Parameter Estimates (selected by default as built-in option and cannot be modified)
  • Residual Sum of Squares
  • Residual Autocorrelation Function
  • Mean Square Error
  • Root Mean Square Error
  • Mean Absolute Error
  • Mean Absolute Percentage Error
  • R Square
  • Adjusted R Square
  • Correlation Matrix of Residuals
  • Eigen Values of Residual Correlation Matrix
  • Test for Normality of Residuals
  • Test for Autocorrelation of Residuals
Plots tab
Select the following options under Plots.
  • Time Series
  • ACF
  • Residual (each variable)
  • Residual Combined
  • Residual ACF (each variable)
  • Fit Values
Save tab
Select Predicted Values and Noise Residuals to save the predicted (fitted) values and residuals of estimated VAR model as additional columns in a new dataset.

Enter a file name and select a location to save the dataset.

Options tab
The following options were selected.
  • Forecast Period: First case after end of estimation period through last case in active dataset
  • User-Missing Values: Treat as invalid
  • Confidence interval Width (%): 95