You may assign "Correlation Attributes" to a Special Event. These attributes can limit the inclusion of historical data that meet a certain condition, or weight the importance of historical statistics used in the forecast. You first define a Special Event to encompass the days and times that the jewelry show airs. This tells Maxima Advantage that historical statistics for these periods of time should be treated differently. You then assign Correlation Attributes that allow the system to discount or augment these statistics when used by the Forecaster. The graph below shows how Maxima Advantage handles your pricing problem.

You can see that the call volume is higher during the shows that featured the lower priced jewelry. You might assign a weight of 1.5 to this show, in anticipation that you will have more calls on reasonably priced items, and most items you sell are reasonably priced. You might then assign a weight of .5 to the show featuring the high priced jewelry. This indicates that Maxima Advantage should lessen the importance of historical statistics received during this show, since you haven't found another Faberge Egg to sell. The forecast will then produce an expected call volume for moderately priced jewelry that is closer to the number of calls received when more reasonably priced jewelry is sold.

Correlation Attributes may also be used to include historical statistics based on a limiting factor. You can assign different attributes to each of your jewelry show hosts. Maxima Advantage can then include only the historical statistics that were produced when a certain host appeared on the show. The graph below shows how Maxima Advantage handles the hosting problem.

You can see that the call volume is higher during the shows that a soap opera star hosts. By assigning limiting Correlation Attributes to both hosts, you may choose which historical statistics are included. When the soap opera star next appears to sell diamond tennis bracelets, you can include historical data received during their other appearances, and exclude historical statistics that were received when an olympic gymnast hosted the show. You can then decide whether to use the star's attribute or the gymnast's attribute when the cable-access weather girl hosts the show.

By assigning both types of Correlation Attributes to a Special Event you have included only the historical data received when a certain host appears on the show, and you may weight these data based upon the sales price of the item. The graph below shows how Maxima Advantage incorporates both types of Correlation Attributes to produce an accurate forecast.

Maxima Advantage did not consider any of the shows that the soap opera star hosted. The Forecaster then used the same pricing weights applied in the first example to only the historical data received during the gymnast's shows to produce a forecast that accounts for both the sales price of the jewelry and the popularity of the new host.