What is Process Mining?
Process mining is a technique enabling the analysis of business processes – as they have been executed – based on so called event logs, which you can extract from information systems. It is somehow comparable to Data Mining, though focused on acquiring Business Process Intelligence.
As you will read, process mining allows to observe the organisational behaviour, based on really occurred facts & figures. A sound basis for in depth investigation and thus for future operational improvement.
What are Event Logs?
Event logs can be seen as collections of traces from cases (also known as “process instance”), resulting from process executions. Assume that following process (which we already used in the blog of 5th of February), regarding the supply of raw materials.
When this process is supported by a software application that registers every transaction of each process instance, we might obtain a (piece of) event log looking like this:
The elements indicated by the green header – i.e. Delivery number (which is the actual “case ID”), Activity and Timestamp – are the 3 mandatory attributes of an event log for process mining. Other additional attributes like User, Product, Weight, Truck ID, etc., may help to enrich process analysis even more, but are not indispensable.
Main applications of process mining
Process Mining basically supports 3 main process management disciplines:
- Business Process discovery: which is the graphical representation – say the reconstruction based on event logs – of an organisation’s current business processes; including possible process variants. This means that you can obtain process models – even process animations – from event logs.
- Conformance checking: this enables the comparison of an existing process model to an event log of the same process, so you can check if reality, as recorded in the log, conforms to the process model and vice-versa. This is convenient to check whether (business) rules have been respected or not.
- Business Process enhancement & improvement: thanks to the graphs and the metrics that you can get from process mining algorithms, it is easier to detect bottlenecks, to see which resources are the most utilised or under-utilised, etc. The many analytical views you can get, facilitate the detection of causes of inefficiencies or errors within a business process.
Main benefits of process mining
Here are the main benefits of Process Mining, according to a survey by J. Claes and G.Poels :
Objectivity: facts never lie; and as event logs reflect how a process (instance) has been executed, process discovery – thus the representation of how processes have really been run – is highly objective. The processes models reconstructed through process mining are not subject to perceptual bias, in contrast to models obtained by interview.
Speed: though you always need a first basic understanding of the business upfront (e.g. to assess which event logs will be relevant to be analysed), it is obvious that reproducing process diagrams through process mining techniques may considerably speed up activities like process mapping and modelling.
Little effort (efficiency): process discovery through process mining enables a considerable increase of efficiency thanks to the fact that you most often need less workshops and interviews of business people.
Completeness: when analysing relevant and consistent event logs, you will quite easily distinguish the mainstream process from other process variants or exceptions. At least, you will be able to spot the variants and exceptions, which you may ignore without process mining.
Transparency: going along with objectivity, process mining also enables to “drill-down to the facts” to obtain details about who executed a transaction, at what time, etc. It helps to get even more transparency on how the organisation really works.
Conformance: like explained previously, process mining allows to identify quite easily non-conformances against the expected or defined process.
Root-causes & bottlenecks: allow to (more) easily find out root causes, bottlenecks, etc.; this obviously eases (continuous) process improvement practices.
Predictions & simulations: Based on facts, it is possible to predict future process behaviour. Event logs also allow to make even more reliable process simulations than the ones purely based on theoretical assumptions. This blog explains in more details how process mining contributes to better process simulations.
Main process mining challenges
Despite the numerous benefits of Process Mining, there are some challenges, however. Claes & Poels mention following “drawbacks”, which I prefer to name “challenges”:
Data access: it is not always possible indeed to get valid event logs. Some ‘older’ applications do even not produce event logs.
Data preparation cost: even if event logs are available, it may need important efforts to obtain those, or to combine logs from several systems to be usable.
Data quality: it may be a challenge to obtain event logs of acceptable / reliable quality as well, depending on the sources.
Intuitive / guidance: not all process mining tools are very user-friendly.
Tool limitations: some tools are more limited than others in the choice of mining algorithms
Hard to understand: for people having only few experience or skills with regards to process mining and tools, it may be less obvious to apply.
Though this first blog on Process Mining may seem somehow theoretical, in next blog I will describe and illustrate the more practical side of it, using scenarios. This way, you will get an (even) better understanding of how you may improve your organisation through Process Mining. Should you have any question or remark meanwhile, then please drop it in below Comment box, or via the contact form.
P.S.: Please don’t hesitate to share this information with your Facebookfriends and fans, LinkedIn contacts, Twitter followers and Google+ circles, through the share buttons below. Thank you!