What is process mining?

It is a data analysis technique based on process. It analyzes event logs which are data recorded in various systems used for work (ERP, CRM, MES etc.). By analyzing the event log data, processes such as the flow of certain work, how a certain product is manufactured and which itinerary a customer goes through within a service, are identified and visualized.
By analyzing the event logs recorded in the system through process mining, the company can transparently and accurately pinpoint the process and furthermore, it is also possible to predict which process a product would undergo to be manufactured or which process a certain customer undergoes when using a service. Through such, it is possible to accurately understand the current business process of the company, along with diagnosis of conformity through comparison between previous process and actual process, re-work, reduction of cost through improvement of waste section, reduction of work time, performance monitoring by work or worker and others.

Process subject to process mining

It can be applied to all (basic, support, supply, use, outsourcing) processes that is managed as a system among internal and external process of the organization.

Company’s customer service handling process

Assessment process of Health Insurance Review & Assessment Service

Hospital’s diagnosis and treatment process

Process for handling government’s subsidy support

Process of handling loan applications in banks

Software development process

HR management process

Process for handling customer service in companies

Health Insurance Review Assessment Process

Hospital Diagnostics and Treatment Process

Government Subsidies Management Process

Bank Loan Application Management Process

HR Management Process

Software Development Process

Looking at the latest trend, more event log information can be collected from various IT systems and more detailed information on the history of process can be received so it has become easy to obtain materials required for process mining. Moreover, under fierce competition and rapidly changing corporate environment, the demand for improvement and support of business process in order to enhance the company’s competitiveness is continuously on the rise. Due to these two reasons, process mining has been adopted and the interest in it is increasing.

Utilization of process data (event log)

In order to apply the process mining technique, extracting “Event log” is compulsory but data items are not required. Majority of the work systems leave the system usage histories in the database or file logs and others, and among them, the tracing subject (case) is determined and as long as “who (executor)”, “when (timestamp)”, “what (activity name)” can be extracted, process mining can be applied immediately.

Here, case refers to the data parameter which occurs equally in the entire flow such as customer ID, registration and purchase number and manufacture allocation number. For instance, even in the case of a system web log that we can see commonly, a process pattern exists, the ID of the user who has logged in is defined as a case and the process map can be drawn concerning the activities carried out in the web system, considering each accessed page as an activity.

Application scope of process mining

Process mining analysis can be applied to all fields where process and data related to it exist. Process mining analysis can be applied comprehensively in various fields such as course of processing cases in public services, manufacturing process collected from various equipment and terminals, analysis of outpatient’s movement in hospitals, treatment and diagnosis pathway and others.

Just like other data analysis techniques, process mining is just a way of supporting decision making in each area and achieving work efficiency and cannot become a target by itself. However, by providing vivid process site images based on data for cooperation with domain experts of each field, it could become a management tool for innovating processes.

Public field
  • Analysis of work performance and identification of bottleneck point by identifying work process model
  • Analysis of work pattern through frequency analysis by position, by day and by time
  • Analysis of association of work among employees/positions by analyzing the use of the System’s menu
Financial field
  • Process improvement through process analysis of loan works
  • Structuralization and reduction of processing time through process analysis of insurance claims
  • Identification of standard process and confirmation of data for BI system and dashboard setting
Manufacturing field
  • Analysis of process compatibility and identification of bottleneck point by identifying process model
  • Performance analysis per process through analysis of manufacture, standby/transportation and time
  • Process schedule management of multiple projects
Hospital field
  • Analysis of root cause of delay in waiting and improvement by analyzing hospital treatment process
  • Analysis of root cause and improvement by analyzing unfinished pattern analysis by examination type and treatment department
Port/logistics field
  • Analysis of movement concerning container flow and identification of inefficient container flow
  • Analysis of loading status and flow by yard
Convention field
  • Analysis of visitor’s movement
  • Analysis of visit time by booth and associated relationship
  • Optimization of booth batch considering movement

Process mining analysis method

Discovery of process
(process visualization and analysis)

Event logs collected from work system and others can automatically draw process maps through process mining analysis. This is the stage of process discovery. The process map that has been identified is abstracted by various algorithms such as alpha, heuristic and purge then converts the process model seeking to find problems into a form that can easily be understood. Other than that, the conformity of work execution can be evaluated from the organization chart’s perspective by schematizing the precedence relationship of work from the process perspective and indicating in a social network form.

Process conformance inspection
(process monitoring and comparison)

The process model identified from event logs can be compared with the standard process that is known previously. At this point, difference between standard process and actual process is expressed in figures by calculating the process model conformance. In addition, through comparison of two processes, unused process which doesn’t actually occur or hidden process which only occurs in actual process can be identified.

Process improvement
(continuous improvement)
The process model obtained from process mining can validate how well the process pattern of various logs given can be explained through the process conformance mentioned previously. Through it, the work of improving process model is carried out so that the occurrence pattern of more log events can be explained.

Types of process mining utilization

The process mining analysis technique can be applied to various fields where process exists. Other than the process map, organization chart and work network analysis mentioned previously, it is utilized for the purpose of analyzing performance within organization by department and by user, improving efficiency factors such as detection of repetitive work and bottleneck section within process and auditing such as compliance of regulation such as detection of abnormal processes. Lastly, it can be used for the purpose of monitoring the execution of process in specific section in the timeline through animation function that allows actual log event to flow into the process map that has been identified.

Types of process mining utilization

As for small data subject to initial analysis, the data collection with small number of single processes (cases) but includes the entire course from the beginning until the end of the process is suitable. Through such small data analysis, the format of data and work subject to key analysis can be defined and the application effect can be verified within a short period of time so it can save time and effort spent in excessive HW infrastructure and analysis in the initial stage.
However, as for the analysis architecture, it is recommended that it should be in an expandable structure from the beginning so that the big data saved in Hadoop, NoSQL, database collected previously can be analyzed at all times using the same logic.