What is Process Mining?
Process Mining is a combination of data mining and process modelling. The process patterns are traced on the basis of the collected log data. There is Process Mining software needed to visualize the patterns and to analyse them. Process Mining a part of data mining where we explicitly search for data which tells us something about the way in which the processes are executed.
Is Process Mining a part of Business Intelligence?
As we gain a better understanding of the performance of processes by using Process Mining, it is also possible to monitor and control the actual outcome. In that case, you can see Process Mining is a part of Business Intelligence. In Business Intelligence the perspective is on numbers. With Process Mining the perspective is on processes.
What is needed for Process Mining?
For the visualization and analysis of the processes you need Process Mining software. There are commercially available software packages, whether in the cloud. In addition, there is an appropriate data set needed which can be read in the Process Mining tool. The dataset can have multiple formats, including CSV, or the data can be uploaded directly from a database.
What is the output of the Process Mining Tool?
Process Mining tools provide a graphical representation of the process. All variants of the process are visible as well as the times and frequencies of the process activities. In addition to the graphical representation of the process there are graphs and other visualisations available for analysing the process.
Why should you use Process Mining?
Process Mining is used for multiple purposes. There may be a need for more insights into the outcome of the process, improving efficiency, more standardized processes, compliance and auditing questions, etc. We recommend that you not consider Process Mining as the main goal. First, it will be part of a project like Continuous Improvement, Auditing programs, Lean, Six Sigma or a standardization program. In a later phase, Process Mining, will be used in Continuous and Real-time monitoring, Continuous auditing and Predictive analysis.
Is Process Monitoring possible?
If it’s technically possible to refresh the data constantly, real-time Process Monitoring is possible. Process Monitoring can also be achieved thru updating the data periodically in e.g. every day or every week. With Process Monitoring you should decide what the best time-interval is between updates of the data. If you apply Process Monitoring, you have also the ability to implement early warning.
Can I comply a process design to the real process?
It is possible to upload a formal process model in Celonis PI or draw a new model. The output of most widely used process modelling tools can be used. The Celonis PI software automatically shows you where the designed process (the as-is situation) deviates from the intended design (the to-be situation).
When is it not possible to use Process Mining?
If a process is not supported by an IT-system, then there is little or no log data available to visualize and analyse the process. In case the process is supported by an IT-system, but there is no logging of data, Process Mining is also not possible.
How can you do it?
It begins with the question. What needs to be improved, or what do you want to know about the real execution of the processes? After determining the scope of the analysis, the necessary data can be extracted from one or more IT-systems. The data is then read into the Process Mining tool and analysed.
Is it difficult to find the appropriate data?
In case you have a process aware information system (like Workflow or BPMS systems) in place, it is not difficult to extract the right data for Process Mining. If you don’t have a process aware information system in place (like ERP systems or a Patient Health Record), it can be difficult to get the right data. We recommend in any case to first make a thorough analysis of the existing systems and their data. When first determining the effort required to have access to the required data, prevents disappointments during the process. In general, extracting and preparing the data can require a number of days. We aim to get the data available in the simplest possible way. Preferably automated or using standardized data extracting queries.
Is there any limitation on the amount of data?
The Process Mining tools are generally able to cope with large volumes of data. It is necessary to determine on forehand the size of the dataset that must be analysed in term of records. It is possible to analyse large amounts of records, from several millions of records until an unlimited amount of records. Immediately beginning with a 10-year history can be an option, but usually it is advisable to take a slice / subset of one or more quarters. And perhaps extend it in a later phase.
Should all data be in one system?
When starting with Process Mining it is preferable to start with a process which is supported by one single IT-system. Over time, you can extend your analysis with data from multiple systems. The data will then be merged. There is no limitation about the source of the data. It can be all log-data from all kinds of IT-systems (Production systems, transaction systems, Websites, mobile devices etc.).
Is it possible to extract the process performance from Process Mining?
Because the use of timestamps in the log data, the Process Mining tool will automatically calculate the through put times and visualize them in the process model. By using colours when displaying the through put times you can immediately see your delays when it comes to processing through put times. In the next picture the colour of the connection (the arrows) has changed in case of a long throughput time.
Are there for process analysis, other data analysis tools necessary?
Depending on the problem it may occur that there is an additional visualization necessary in order to obtain a particular insight. The Process Mining software itself is developing rapidly and will integrate the necessary analysis technology within the Process Mining software e.g. ‘R’ for predictive analysis and visualisations for Social Network analysis.