Chapter 11 Times Series 445
Figure 11-9
ACF for the
change in
the annual
temperature
5 Enter 20 in the Calculate ACF up through the lag spin box.
6 Click the Output button, and send the output to a new sheet named
Change ACF. Click OK twice.
Figure 11-9 shows the output from the command.
The autocorrelations for change in average temperature are not as strong
as you saw earlier using the yearly temperature values. However note that
the lag 1 and lag 2 correlations are both statistically signifi cant and negative.
This indicates a negative correlation between the current change in tem-
perature and changes from one or two years prior. Apparently an increase
in temperature in one year is associated with a smaller increase or even a
decrease in the next two years.
Moving Averages
As you saw earlier in Figure 11-5, the change in average temperature can
vary unpredictably from one year to another. One way of smoothing out this
fl uctuation is to take the average change over an entire decade as you did
for the yearly temperature values. Another way of smoothing your data is to
calculate a moving average.
