Chapter 11 Times Series 471
The autocorrelation holds between adjacent months but the highest
autocorrelation exists for sales that are 12 months or one year apart. Also
note a second signifi cant autocorrelation occurs at 24 months. So the ACF
results do show a seasonal correlation of the sales fi gures. In other words,
the pattern of sales from one year to the next is fairly consistent.
Adjusting for Seasonality
Because the liquor sales data have a seasonal component, it would be useful
to adjust the values for the seasonal effect. In this way you can determine
whether a drop in production during one month is due to seasonal effects
or is a true decline. Adjusting the production data for seasonality also gives
you a better indication of the trend in liquor sales over the course of the
10 years of the study. You can use StatPlus to adjust time series data for
multiplicative seasonality.
To adjust the liquor sales data:
1 Return to the Liquor Sales worksheet.
2 Click Time series from the StatPlus menu and then click Seasonal
Adjustment names list. Click OK.
3 Verify that a period of length 12 is entered into the Length of Period
box.
4 Click the Output button and send the output to a new worksheet
named Adjusted Sales. Click OK.
Your completed dialog box should look like Figure 11-28.
Figure 11-28
The Perform
Seasonal
Adjustment
dialog box
