The Five Rules of Process Mining

The Five Rules of Process Mining

The Five Rules of Process Mining

Process Mining merges process modeling and data mining to explain how systems, processes, users, context, and data operate together. It is a specialized branch of data science that delivers more and more insights to corporates.

Rule #1 Business First

Business first

As a general rule, we need to put business first, technology second. Process mining is no exception. This fantastic set of tools is helpful to answer a business question using data.

On a relatively straightforward process, the number of analyses is virtually unlimited. Fortunately, process mining can also help to discover the question. A well-rounded question with a real business value is necessary to start.

Type of Questions

Process mining can answer a vast range of questions:

  • What are the bottlenecks in a process?
  • What are the predictors of the processing time?
  • Which resources are on the critical path?
  • What are the predictors of loops?
  • What are the idle resources?
  • What are the predictors of refusal?
  • Which steps contribute most to the idle time?

Rule #2 Users Lie

Once the business question is defined, a set of interviews on the critical steps of the processes, the business rules, the data manipulated, the controls, and the exceptions. It is a practical first step to draw a very high-level map of the domain, but the first analysis cycles reveal how much these declarative pieces of information are incomplete, inaccurate, and sometimes plain wrong.

There are many biases in what users believe they do versus what they do. Process mining will set the record straight and analyze what is actually happening.

Rule #3 Data is Never Perfect

Input Data

Process mining relies on data that is capture digitally through the processing of a case. “Case” is a general term to define a request, a transaction, that has a multi-step process attached to it. Each case has its own story, as users carry out activities to bring it from the is start state to its end state. Each activity happens at a given time, represented by a timestamp. These are the abstract concepts of the minimum pieces of data to feed into a process mining project.

Data Quality

For the process mining to deliver good quality output, we need good quality input. Unfortunately, data is rarely at the quality required, in the correct format. There are many quality issues to overcome:

  • missing data,
  • corrupted data,
  • inconsistent data,

Pre-processing

The first step of a process mining project is a phase of data cleaning. This phase is made more challenging by Rule #1 because inconsistency between the user description and the data may not be due to data issues. The solution is to rely on consistency within the data itself, between different sources of data, and the usage of correlation metrics.

Rule #4 Simple is Beautiful

Perfect Process

The ideal process is often challenging to design. The user description, the overall business objectives, the context of execution all need to reach a compromise. Aiming at a perfect process without any room for variation is dangerous. A process must leave room for decision and sometimes exceptional handling from the user.

Process Mining in Action

The first tools of process mining are the discovery algorithms. Based on the data extracted from your management system, these tools can automatically draw the process that users follow in real life. At this step, we want to compare reality with theory and assess the gap.

Process metrics

The following four metrics capture the complexity and the necessity for balance in process mining:

  • Fitness: how perfectly the process model fits with reality.
  • Precision: how the model defines this process and only this process.
  • Simplicity: how simple is the process.
  • Generalization: what are the chances that this process covers valid instances that were not present in the data.

Rule # 5 Let the Data Tell the Story

This rule is valid in every branch of data science. The purpose of any data-driven analysis is to let the data tell the story. It is critical to reaching a conclusion that is as void of bias as possible. Each step of a process mining project opens a few new hypotheses to be validated by the following analysis cycle until the data tells the complete story.

We are Here to Help

At System in Motion, we are committed to building long-term solutions and solid foundations for your Information System. We can help you optimize your Information System, generating value for your business. Contact us for any inquiry.

You should also read

Can you Reconcile ERP and Agility?

Can you Reconcile ERP and Agility?

Article 4 minutes read
How Process Mining Facilitates Operational Efficiency?

How Process Mining Facilitates Operational Efficiency?

Article 4 minutes read

Let's start and accelerate your digitalization

One project at a time, we can start your digitalization today, by building the foundation of your future strength.

Book a Meeting