Query order states

Query-order states allows the states of a variable to be automatically re-ordered based on a current query, whilst maintaining any affected distributions.

For example, this can be useful when you want to order the states in a variable from highest to lowest probability.

After learning a cluster model, the cluster states are in no particular order. Often you want the clusters to be listed in decreasing probability (support). This can be achieved using Query-order states, without affecting model outputs.

How to query-order states

  1. Ensure that you have queried the variable you wish to re-order.
  2. Ensure that the node containing the variable you wish to re-order is selected, and click the Query-order states button on the main ribbon toolbar, Network tab, Refactoring group. This will launch the Re-order states from query window.
  3. Select the query to base the sort on. Usually there is only once choice, but their can be multiple choices, for example when the variable is temporal. By default the sort is descending, but can be changed to ascending by setting the Ascending checkbox to true.
  4. Click OK to complete the re-ordering. All distributions that are affected by the re-order (such as child nodes) are updated.