Selecting the Forecast Objective

  To define a forecast objective and, optionally, define requirements:

1. If you have more than one workbook open, use the Primary workbook list to select the workbook with data to optimize.

2. Click Add Objective.

   A default objective is displayed in the Objectives area.

3. Review the default objective definition. It has the format Operation, Statistic, Forecast.

   1. First, if the model has more than one forecast, does the default objective include the same forecast you want to include in the objective? If not, click the underlined forecast and replace it with your selection. If more than ten forecasts are available, More Forecasts is displayed at the bottom of the list. You can select it to display a forecast selection dialog.

   2. Next, do you want to maximize a statistic for that forecast? If you would prefer to minimize the statistic or set it to a target value, click the underlined operation and select an alternative.

   3. Finally, is the underlined statistic the one you prefer to use. If not, click it and select a different one. If you have activated Crystal Ball’s process capability features and have defined an LSL or USL, the process capability statistics are available in the list of statistics.

      Note:

      For many problems, the mean (expected value) of the forecast is the most appropriate statistic to optimize, but it need not always be. For example, investors who want to maximize the upside potential of their portfolios may want to use the 90th or 95th percentile as the objective. The results would be solutions that have the highest likelihood of achieving the largest possible returns. Similarly, to minimize the downside potential of the portfolio, they may use the 5th or 10th percentile as the objective to minimize the possibility of large losses. You can use other statistics to realize different objectives. See the Glossary, online help, and the online Oracle Crystal Ball Statistical Guide for a description of all available statistics.

4. Optional: Define requirements.

   1. To add a requirement, click Add Requirement. A default requirement is displayed.

   2. First, look at the default statistic. Is it the one you want to use? To review the list of available choices, click the underlined statistic and select a different one if you want. Depending on your choice, the requirement statement could change.

   3. Next, review the forecast. If you want, click the underlined forecast and select another.

   4. Then, review the requirement operator. The selected statistic can be less than or equal to a selected value, greater than or equal to a selected value, or between two selected values (including the values). Click the underlined limit to select another. If you select Between, an additional target value is displayed.

   5. Finally, review and adjust the target value or values. To change a value, click it and then type a new number over it.

   6. You can repeat steps3a through 3e to add additional requirements. New requirements are duplicates of the last one entered.

   7. Optional: If you want to set variable bounds for Efficient Frontier analysis, select a variable and click Efficient Frontier. For details, see Efficient Frontier Analysis.

   Note:

   You can create multiple requirements without using all of them at once. If you select Exclude, that requirement is not used in the current OptQuest optimization.

5. Optional: If you have an .opt file from an earlier version of OptQuest, click Import to open the file for assistance in defining new objectives, requirements, and constraints. For details, see Transferring Settings from .opt Files.

6. Optional: To delete a requirement, click it and then click Delete.

7. When objective and requirement settings are complete, click Next.

   The Decision Variables panel opens.