Microsoft Office Tutorials and References
In Depth Information
Exhibit 3.20 Product E quarterly and yearly total data
clearly no clusters of data representing distinct quarters that are easily identiﬁable.
For example, there is only 1 value that falls into the category (bin) of values between
5 and 15 , and that is the 2nd quarter of year 1. Similarly, there are 3 data values that
fall into the 75 to 85 bin: quarters 4 of year 1, quarter 4 of year 3, and quarter 3
of year 4. It may be possible to adjust the bins to capture clusters more effectively,
but that is not the case for these data values. But don’t despair; we still have other
graphical tools that will prove useful.
Exhibit 3.20 is a graph that explicitly considers the quarterly position of data by
dividing the time series into 4 quarterly sub-series for product E. See Exhibit 3.21
for the data selected to create the graph. It is the same as Table 3.3. From Exhibit
3.20, it is evident that all the product E time series over six years display impor-
tant data behavior: the 4th quarter in all years is the largest sales value, followed
by quarters 3, 1, and 2. Note that the Yearly Total is increasing consistently over
time (measured on the vertical scale on the right-Yrly Totals), as are all other series
except for quarter 4, which has a minor reduction in year 3. This suggests that there
is a seasonal effect related to our data, as well as a consistent trend for all series.
It may be wise to reserve judgment on quarterly sales behavior in the future, but
clearly these are interesting questions to pursue with more advanced techniques.
Before we proceed, let us take stock of what the graphical data analysis has
revealed about product E:
1) We have assumed that it is convenient to think in terms of these data having three
components—a base level, seasonality effects, and a linear trend.