QPR ProcessAnalyzer BPMN Editor

This page describes components that are used for editing BPMN models and viewing conformance analysis results. Note that for dashboards that use BPMN and conformance functionality, you need to create a variable designModel with behavior Stored (in the Dashboard Properties in Variables tab), so that the dashboard is able to store the defined BPMN model. Note also that the BPMN editor only allows to add BPMN elements that are supported by the conformance checking functionality.

BPMN Editor

The BPMN Editor analyzes process mining data with the help of a designed BPMN process model. You can for example:

• See process mining KPI's and statistics calculated on the fly top of the BPMN diagram during the modeling the BPMN
• Run conformance analysis using the designed BPMN model
• Filter data directly from the BPMN diagram

The BPMN diagram can be drawn from scratch or imported from a BPMN 2.0 XML file. The ready diagram can be exported to a BPMN 2.0 XML file.

BPMN diagrams are created by dragging items from the tool palette. Start event, end event, task and gateways are the most used elements. When you add a BPMN task to the canvas, all event types are shown in a list next to the task, and an event type can be selected to match it with the created task. When the task is added, KPI values are calculated and shown in the blue background. When flows are added, KPI values are calculated and shown for them, too. By clicking the KPI in the top right legend, the KPI value is hidden, to make the BPMN diagram more readable if there are lots of items.

When you select a BPMN task, it suggest to create a filter. In the dropdown list you can select which type of filter is created. When you select a BPMN flow, it also suggest to create a filter.

You can create a filter to get only the conforming or nonconforming cases based on the current BPMN model by selecting Include conforming cases or Include nonconforming cases from the context menu.

Auto-creating BPMN Diagram from Eventlog

BPMN editor's context menu has the Auto-create diagram option which will automatically create and layout a fully conforming BPMN diagram from the filtered eventlog. The created diagram contains paths for all the process variations in the filtered eventlog. The diagram will have one start element and one end element, all the event types appear as BPMN tasks in the diagram. The auto-creation intelligently identifies the types of gateways, using either the exclusive or parallel gateways. Parallel gateways indicate that there are activities occurring simultaneously in the process that are typically independent of each other.

Eventlogs often contain a large number of process variations, even in smaller datasets. Capturing every variation in a single BPMN diagram can quickly lead to a model that is overly complex and difficult to interpret. A practical approach is to first filter the eventlog to include only the cases that represent the desired variations or process behaviors. Using the variation filter is typically the easiest way to select the process behavior you want to include to the BPMN diagram. You may need to iterate a few times by adjusting filters and regenerating the diagram to arrive at a BPMN model that represents the desired process flow. If the auto-created diagram still does not fully meet your needs, you can always continue refining it manually using the BPMN editor.

The auto-creation feature is available for the Snowflake models, and it uses the expression language's ToBpmn function which generates the BPMN diagram in the Snowflake by calling a Python function powered by the PM4Py library (https://processintelligence.solutions/pm4py). The function is executed in the same Snowflake ODBC connection where the eventlog data is located. To use the auto-creation feature, the required Python libraries need to be deployed to Snowflake as instructed here.

Adding tasks and flows, and Deleting Diagram

There are following functions available:

• Add tasks and flows will create a BPMN model that is the same as the as-is model. Thus, the as-is model will conform perfectly to the design model. After the tasks and flows are created, you can edit the design model and remove the events and flows that shouldn't occur in the process. Note however, that the amount of flows that is created can easily be overwhelming. To overcome this, you can first filter out those events and flows that shouldn't occur in the desired process and then use the auto-create. After creating the design model, you can remove the filter to check the conformance of the whole model. Note that due to performance reasons, the number of flows in the auto-generated model is limited to 1000.
• Add tasks only will generate only the tasks without any flows. The result is a model that needs to be filled with flows to be a valid BPMN model.
• Delete diagram will delete the current diagram and start from an empty canvas.

Reports Selectable in BPMN Diagram

There are also following reports available for in-memory models:

• Duration distribution between tasks (select 2 tasks)
• Flow duration distribution (select 1 flow)
• Task occurrences across time (select 1 task)
• Task repetition in cases (select 1 task)
• You can export the BPMN diagram contents as a flat list by clicking the Export all BPMN elements as CSV.
• You can export a list of BPMN elements in the order that they appear in the flow by clicking the Export element order.

KPI's on BPMN Diagram

The following KPI's are shown:

• Cases having task: Number of cases in the process mining data, that have this task (=event type).
• Task occurred: Number of times this type of event has occurred in the process mining data. This number is the same or larger than the Cases having task count. If it's larger, the event has occurred more than once in a case (looping, repetition).

Flow has a different meaning in the BPMN diagram than in the traditional process mining diagram. In the process mining diagram, a flow is direct transition between two events. This means that no other event has occurred between the start and end event of the flow. In the BPMN diagram. Following KPI's for flows are shown for in-memory models (there are no flow KPI's for Snowflake models):

• Flow median duration: Median duration from the task event of the flow to the end event of the flow.
• Flow occurred: Number of times the flow has occurred in all cases. The BPMN flow occurs when in the case there occurs the starting event and that is followed by the end event at any point in the case.
• Other events between: Number of events that has occurred between the starting and ending event in the flow.

Design Model Selector

In addition to the BPMN Editor, there is also a Design Model Selector component available, which adds a button to the dashboard for opening an overlay BPMN editor. The Design Model Selector is better when there is no space in the dashboard to show the BPMN model.

The Design Model Selector validates the BPMN model when the editor is closed, so invalid BPMN models cannot be saved.

Conformance Statistics

Conformance Statistics visualization shows an overview of how many cases are conformant and deviating (both as absolute case count and percentage of total cases). In addition, the Conformance Statistics component shown average duration of cases and average number of events per case in both the conformant and deviating cases groups. The used BPMN model is stored in the designModel variable in the context of the conformance statistics visualization.