IEEE CIS Task Force on Process Mining

Trace: slide9


XES Webinar

Slide 9: Example

[TAB] Based on event data from such a purchase system, a data scientist could do several things.

As a first example, the data scientist can [TAB]discover a [TAB]process model from the event data, without using any a-priori information. If the event data also contains information about resources, a data scientist can also discover resource-related models, like a [TAB]social network showing how people work together in the organization.[TAB]

Second, on either an existing or a discovered model, the data scientist can [TAB]analyze the performance of the system. For example, [TAB]bottlenecks can be detected, information on flow time can be provided, and the average time it takes to move from one activity to another can be reported.[TAB]

Third, the data scientist can [TAB]enhance either an existing or a discovered model. For example, [TAB]information on where models deviate from an event log can be projected onto the model, frequent execution paths could be visualized, information on execution and waiting times could be provided for activities, and people who often hand over work to each other could be shown.[TAB]

Finally, the data scientist can [TAB]provide prediction diagnostics on the current process. For example, he can report [TAB]which activities occurred in situations where they were not supposed to occur according to the model.[TAB]