How to use the process optimization tool

Use the process optimization tool in a flow rule to perform probability analysis on assignments to identify common flow patterns and trends. See About the process optimization tool for a description of what it does and how you implement it.

To optimize assignments in a flow, perform the following steps.

Step 1. Define the properties in each assignment

Open a flow rule.
Right-click the Assignment shape(s) you want to optimize, which opens a pop-up window.
Select the Optimization tab.
Select one or more a valid Single Value properties in your work class.
Click OK. The window closes.
Optionally, select another assignment and repeat the above steps.
When you are finished, save the flow rule.

The properties are saved with the flow rule and are used in subsequent optimization sessions. You can override these settings for a single optimization run using the Optimization Criteria panel as described in Step 2.

The optimization process includes only work items that have values for all the properties listed in the array. Incomplete records are ignored.

Performance guidelines:

In general, numeric properties require more processing than text strings.
Performance is enhanced by the use of regular, repeating data. For instance, policy types or city names process faster than customer names or addresses.
The number of possible values for a property directly affects performance — the fewer the values, the better the performance.

Step 2. Define the optimization criteria

Select Optimization from the Overlay drop-down list. This opens the Optimization Criteria panel, which contains the following options.

The settings return to the defaults after an optimization run.

Field	Description
Filter	Select one of the following: Date — Filters records within a date range you specify in the From Date and To Date fields. Analysis is performed on the properties you specified in unless overridden in the Advanced section. This is the default. Custom — Optional. Filters records using the criteria defined in a list view rule. When you select this option, the List View and List View Owner fields replace the From Date and To Date fields. See the section Using a list view rule in this article.
From Date and To Date	Enter dates to define the range of work item records. Dates are determined by work item created values (`pxTimeCreated)`. By default, the range is one month as of the current date.
List View	Select the Purpose key part of the list view rule for the work type.
List View Owner	Enter the Owner key part of the list view rule for the work type. The default value is ALL.
Maximum Results	Enter an integer indicating the maximum number of records (as identified by the object property pxFlowAction) used in the process. No limit is enforced. The count begins with the most current record. The larger the number, the greater the memory requirements and processing time. The default value is 500.
Run in Background	Select this checkbox if you want to work in PRPC during the optimization process. By default, the checkbox is not selected.
Only show results with n % of certainty	Enter a value (as a percentage) between 0 to 100 that filters the statistics displayed in the Optimization Results panel. For example, a value of 85 displays only the results that range from 100% to 85% certainty. The optimization algorithm accuracy as determined in Thoroughness (%) field. The default value is 85.
Advanced Options
Thoroughness (%)	Optional. Enter a value (as a percentage) between 0 to 100 that defines the scope and precision of the optimization process. For example, a value of 85 produces results that range from 100% to 85% in accuracy — a 15% error rate. The higher the threshold, the longer the processing time. The default value is 85. As a best practice, use the same value in the Only show results... setting. An error rate that is higher than your threshold is possible. This occurs when there is not enough data for the wizard to reach a valid conclusion. For example, if you run the Process Optimization wizard with a single property that has discrete values, such as .ColorName, the decision tree can only be one level deep (.ColorName = "blue", .Color = "green", and so on). In this case, a flow action for blue may account for 70% and another30% for green. These results may not be accurate enough to meet the threshold. The default value is 85.
Assignment name	Optional. Select an assignment if you want to override the properties defined on its Optimization tab. The values in the tab are not used in this run.
Optimization Properties	Select one or more properties if you have selected an Assignment name value.

Step 3. Run the optimizer and display the results

On the Optimization Criteria panel, select Optimize. When the process is complete in Process Modeler flows, the optimization results appear in a window when you hover over the icon.
Note that if a property has no value, it appears with empty quotation marks in the panel and in subsequently generated decision trees or decision tables.
Optionally, select either the Decision Tree or the Decision Table links to create a rule using the statistics. Here is an example of a decision tree:

Using a list view rule

You can use a list view rule to filter the results based on the properties and date range defined in the rule. Do the following:

Create a list view rule in your work class with an Owner value (such as ALL).
On the Display Fields tab specify the following four history properties and at least one existing class property. The prefix used in this example (historyPage) is consistent with the prefix name specified on the Join tab as described in Step 9.

Field

historyPage.pxFlowAction

historyPage.pyFlowType

historyPage.pyTaskID

historyPage.pxHistoryForReference

.Category

On the Content tab in the Criteria section enter the following:

Field	Condition	Value
historyPage.pxFlowAction	Is Not Null	<blank>
historyPage.pyFlowType	Is Equal	param.flowType
historyPage.pyTaskID	Is Equal	param.assignmentID
historyPage.pxHistoryForReference	Is Equal	.pzInsKey

Specify And logic for all rows: A AND B AND C AND D.
In the Report Source section, enter OptimizationRecordsPage in the Content Page Name field.
On the Organize tab, clear the Enable Paging checkbox in the Paging section. Also, clear the Custom View checkbox in the Organize Title section.
On the Parameters tab, enter two required parameters: flowType and assignmentID.
On the Pages & Classes tab, enter the following in the Criteria section:

Page Name	Class
	PegaProjMgmgt-Work
historyPage	History-PegaProjMgmgt-Work
OptimizationRecordsPage	Code-Pega-List

This historyPage page name is consistent with the prefix name specified on the Join tab as described in Step 9. It must be in the history work class of the flow (PegaProjMgmgt_Work in this example).

The OptimizationRecordsPage name is consistent with the content page name specified in the Content Page Name on the Content tab. This page must be in Code-Pega-List class.

On the Join tab, enter the history page name and history class in the Prefix and Class Name fields (historyPage and History-PegaProjMgmgt-Work in this example).
Click Edit Conditions to open the Join Conditions dialog.
Enter the property historyPage.pxHistoryForReference, a condition of Is Equal, and a value of .pzInsKey.
Click Apply and save the rule.

	decision shape
	About Decision Trees Property-Map-DecisionTree method
	Atlas — Standard Decision Trees

How to use the process optimization tool

Field

Using a list view rule

Condition

Value

Class