Case variants
The Case Variants panel shows the event log statistics, such as frequency, throughput, and the number of events, aggregated by the following dimensions:
- Case Structure (variants)
- Activity
- Resource (if available)
- Role (if available)
- Custom fields
Analyze variants
By selecting Case Structure, it is possible to find the most frequent variant and compare it with other variants in terms of performance (average lead time), number of events (flat processes) or relations (multilevel processes).
For each variant you can view the following information:
-
Frequency:
-
Flat processes:
The number of cases that are related to that variant
-
Multilevel processes:
The number of composed multilevel end-to-end cases that are related to that variant
-
-
Number of events
In the multilevel processes, even if the number of events is different for the two composed cases, the number of relations can be the same. Therefore, the number of relations must characterize a variant. For example, an invoice that refers to multiple orders can follow the same variant of another invoice that refers to only one order.
Important: For the Resource, Role, and custom fields, the
number of viewer events
value in the Case Variants panel corresponds to thenumber of distinct activities
value on which the attribute is valued. -
Throughput (Average Lead Time)
You can sort variants in either ascending or descending order using the
Frequency
orThroughput
value.
In the following example:
- 50% of the instances follow the most frequent Path
- The Path is composed of
10 events
- The Average lead time is
90 days
and3 hours
By selecting the Zoom button (active only for Case Structure), IBM Process Mining provides information about the average waiting time, average service time, and frequency of the activities involved within the selected Case Structure or variant.
Filter by variants
You select one or more variants and choose to include or exclude the matching cases.
You use the following steps to create a new filter with variants:
- Select a variant.
- Press Ctrl (on Windows or Linux) or command (on mac) and select all the variants you want to filter.
- Click Filter to see the process.
- Create a user-friendly name.
- You see the new filter with the assigned name.
Filter by activity or custom fields
By selecting an Activity, Role, Resource or custom field in the drop-down menu, it is possible to group the variants by the value you selected in the field.
IBM Process Mining identifies the cases in which an Activity, Role, Resource or custom field is involved at least one time and the related paths, as shown in the following example:
The statistics represents:
-
Frequency:
Percentage and number of cases in which the activity or attribute is present.
-
Number of events:
Average length (number of events) of cases in which the activity or attribute is present.
-
Throughput:
Average lead time of cases in which the activity/attribute is present.
When you click Filter, IBM Process Mining redraws the process by using only those instances.
Important: Filtering for a certain field’s value does not necessarily exclude the other values of the field. For example, if a model is filtered by considering a specific resource, other resources can still appear after the filtering. This is because the selected resource might be involved in cases in which other resources are involved (the filtering is on the process cases).
Levels of attributes
Based on the analyzed process and on how datasource were prepared, there may be 3 different levels of attributes in IBM Process Mining:
-
Case-level
attribute:A
case-level
attribute is an attribute with uniform value across a whole case (either flat or multilevel). It has a uniform value on all the events belonging to the same case. Examples of thecase-level
attributes arecase type
,country
, andcompany
.- To analyze the distribution of the
case-level
attributes, use the "Case variants" panel.
- To analyze the distribution of the
-
Entity-level
attribute:For the multilevel processes, an
entity-level
attribute is an attribute with uniform value across the wholeentityID
parameter, but not across the whole multilevel case. Typically theentity-level
attributes are blank on events not belonging to the respective entity. Examples ofentity-level
attributes areorder price
,invoice type
.- To analyze the distribution of the
entity-level
attributes, enter the "Statistics" panel of the respective bridge activity.
- To analyze the distribution of the
-
Event-level
attribute:An
event-level
attribute is an attribute that may change in every event, even inside the same case. Examples ofevent-level
attributes areresource
,role
,activity type
.- To analyze the distribution of the
event-level
attributes, enter the "Statistics" panel of the activity you are interested in. In case you are interested in the number of cases a certainevent-level
attribute appears, use the "Case variants" panel.
- To analyze the distribution of the
Important: If a case-level
or entity-level
attribute changes while running the case (for example, a change in the case type
attribute). This
creates 2 values of the same attribute inside the same case and may affect the statistics and the filtering activity.