Digital Transformation

How Process Mining Finds Hidden Business Processes

Note: This article was originally published on my previous blog ProcessRamblings.com.

In this article, we are going to discuss what type of gaps are left by existing process analysis techniques, what is process mining and what are the benefits of applying process mining techniques during analysis and how to apply these techniques.

Gaps In Process Analysis Techniques

First, let’s take a look at the cases in which existing process analysis techniques fail and result in process gaps.

The picture on right is taken from the process mining course that is offered by Eindhoven University of Technology. It gives insight into an important human behavior of

Shortcuts.

Shortcuts are very common in our daily lives. As the picture shows there might be a perfectly laid out path in front of us, but most of us will still create our own shortcut through the grass.

The same practices can be seen in our work lives. An organization can have very well defined processes, but workers still find a way to bypass them. These bypasses are usually the most commonly executed flows of the process (a.k.a. Process Highways), yet in most cases remain undocumented.

During process analysis, requirements are usually captured from the following sources:

• Workers/Performers
• Standard Operating Procedures (SOP)
• Process Documentation
• Industry Standards

All these sources provide information on the as-is process i.e. how it has been defined or how the process is supposed to flow. This does not necessarily reflect the reality and as a result, we find gaps between the defined process and the executed process.

Defined versus Executed Process

To further elaborate this gap, consider the example of a simple Expense Claims Process. The first process model shows what is the defined flow of an expense claim request. The assumption is, that 100% of claims after submission go for Manager’s approval. Manager approves or rejects them, and in the case of approval, they move to the Finance team’s queue for reimbursement.

In reality, workers do not follow the defined process. If we were to look at the historical data of the process, we might find that in reality only 30% of claims are sent for Manager’s approvals. 70% of claims are directly sent to Finance team for reimbursement.

We have established that people bypass defined processes and they create their own shortcuts, yet most of the process analysis techniques we use only look at the defined as-is processes and simply ignore these bypasses. So this results in gaps between the defined process and the executed process.

What is Process Mining?

In the expense claims process example what actually helped us in identifying the gap?

Data

This is exactly what process mining techniques do. They allow us to extract real process information from the data. These techniques identify all possible paths of the process from events log and answer following questions:

• How are the cases actually being executed?
• What is the most frequent path for every process model?
• How is the distribution of all cases over the different paths through the process?
• How many people are involved in a case?

All these questions, help us in extracting the process paths that are actually executed in the organization.

How To User Process Mining Tools?

There are a few process mining tools available on the market. One of them is ProM, which is a free tool developed by Eindhoven University of Technology in collaboration with a few other universities and vendors.

At a high-level the usage of ProM is simple.

Step 1 – Retrieve events log of the process that needs to be analyzed. Events log usually has following information that can help identify transactions uniquely and correlate them.

Case ID  |  Activity Name  |  Performer  | Timestamp

Step 2 – Events log are converted into an XML format that is understood by ProM.

Step 3 – ProM mines the data and generates all possible process paths using the standard BPMN. The generated model includes detailed information including all possible process paths, participants and number of times each flow is executed etc.

Conclusion

In my opinion, process mining techniques can be extremely useful in various different scenarios, such as:

• Process analysts can use these techniques to verify process models i.e. to ensure that all possible process paths and participants have been captured. These techniques can also be very valuable at the analysis and optimization stages.
• As more and more organizations make Internet of Things (IoT) an integral part of their processes (see auto insurance industry, waste management industry), we will need to mine data generated from these IoT devices. This data will help in understanding how these IoT devices create their own process highways.

Currently, the tools available for process mining are mostly standalone and have a relatively higher learning curve, as a result, there might be some reluctance on their usage. Even if you are not planning to mine data using the tools, the techniques themselves can be used. You can use the events log data to manually run all possible scenarios to make sure that the process model can handle real life situations.

Business Process Management Software (BPMS) vendors do provide simulation features, but they only work on historical data generated by the BPMS. So BPMS vendors need to add process mining features in their offerings as well (either by building from scratch or acquiring and integrating the already available tools).

If you are more interested in learning about process mining techniques then visit the links provided in the references section.

References

• Process Mining: Data science in Action
• ProM 6 Tutorial
• Discovering, Analyzing and Enhancing BPMN Models Using ProM
• Fluxicon – Webinar introducing a commercial process mining tool based on ProM

by Adeel Javed