Similarity Calculation

Similarity in demand can be determined using either transaction-based data or attribute data. You can calculate the similarity using each type of data, if both types are available. You can only view the most recent run in the UI, so in order to compare runs, you must query the database to obtain the results from earlier runs.

Each category has its own set of similarities, relevant to the SKUs that are in that category. A similarity is calculated for each pair of historical SKUs in a category.

If transaction-based similarities are available, it is recommended that you use them instead of attribute-based similarities. Note that transaction-based similarities are only available through the Customer Decision Tree application.

Process

Here is the high-level process for calculating similarities:

1. Enter a unique name for the version of the DT model to be calculated.

2. Select the source of the data to be used: transaction-based or attributed-based. Transaction-based data is only available from the Customer Decision Tree generation.

3. Use the check boxes to indicate whether or not only top-level processing should occur for location or customer segment.

4. Customize the ranking of the category attributes if necessary.

5. Click Run to start the calculation.

6. Review the calculation results in the Similarity Display.

7. After reviewing the results, if necessary, make changes to the values for the calculation in the Version Setup and Category Attribute Setup and then re-run the stage.

8. When you are satisfied with the results, click Next to go to the Elasticity Calculation stage.