How open-source software is helping business improve its processes
The IT systems that support our modern business operations record vast amounts of valuable data daily. Process mining gives organisations the keys to this data, providing analysis and insight into making more strategic, data-driven business decisions.
Through large-scale collaborative research incorporating multiple research institutions and industry partners, we’re investigating how we can bring the full range of process mining elements together to create a practical and powerful solution for helping organisations improve their performance and support their digital transformation journey.
What is process mining?
Process mining is a family of techniques used to analyse the performance and conformance of business processes, based on transactional data recorded by common IT systems within an organisation. If this data remains untapped, it can result in a stark contrast between the perception and reality of how business operations are conducted, resulting in ill-informed decisions.
These techniques complement tactical business intelligence dashboards however, whereas these dashboards offer managers and analysts an aggregate picture of the health of a process, we’ve found that process mining techniques allow us to dig deeper. This deeper investigation allows us to “discover” exactly what tasks are performed and the order in which they occur.
Breaking down the performance of a process to the level of individual tasks, resources and hand-offs also let us identify bottlenecks that may cause delays. Techniques also include differentiating between process variants – for example, users in the banking sector exploring why certain low-value loan applications take longer than others to be processed.
Process mining is a family of techniques used to analyse the performance and conformance of business processes
We can characterise deviations from policies and regulations and gain an understanding of the root causes of such deviations, such as identifying how often invoices are released without being approved and thereby developing an understanding of why this occurs.
Armed with these insights we can also become proactive, predicting future process performance and outcomes as process cases unfold. This could include predicting whether a claim will be handled within the time-frame specified by a service level agreement, whether a loan assessment will be positive, or even whether a loan offer will be accepted by the applicant.
The plugin for discovering process models and mining process performance from event logs
The result of 10 years of collaborative development, Apromore is the first open-source tool to implement the full spectrum of this process mining functionality, from automated process discovery through to performance mining, conformance checking, variants analysis and predictive monitoring.
The project is led by a team of University of Melbourne researchers comprising myself in addition to:
- Abel Armas-Cervantes
- Adriano Augusto
- Raffaele Conforti
- Hoang Nguyen
- Alireza Ostovar
- Artem Polyvyanyy
- Simon Raboczi
- Ilya Verenich
As well as this core team, the project includes contributions from Estonia’s University of Tartu (led by Marlon Dumas), Italy’s University of Camerino (led by Andrea Polini), Queensland University of Technology, The Netherlands’ Eindhoven University of Technology or TU/e, CSIRO’s Data61 (Nick van Beest), Germany’s Co-operative State University Karlsruhe (Thomas Freytag) and more.
Apromore in practice
Apromore has been used with organisations both domestically and internationally across a range of sectors including banking and insurance, IP management, higher education, IT service desk and healthcare.
Our first project with Banca Intesa Sanpaolo, Italy’s largest banking group, used Apromore to gain insights into deviations from the standard way of managing deals in foreign exchange trading processes. A second project will focus on predictive and prescriptive business process monitoring.
In a nutshell, we plan to build on our analysis of historical data (offline), aimed at supporting business analysts in identifying process improvement opportunities, and then move onto the real-time analysis of event streams (online). The idea is to equip operational managers with a range of predictive analytics on a process case basis; i.e. predicting whether a loan offer will be rejected because it is taking too long to assess.
These predictions can be used to recommend a set of interventions for managers, driving process cases towards higher levels of performance and conformance. This may mean shifting resources from one loan on to another to ensure the latter is assessed in a timely manner.
We plan to carry out a follow-up project along the same lines with Unipol, Italy’s largest insurance group.
How collaboration is driving real-world outcomes
With Apromore, we’ve explored two types of partnerships to deliver research outcomes benefiting both the business and research communities.
The first involves partnering with a consultancy firm, such as P4I in the case of Banca Intesa Sanpaolo or HSPI in the case of Unipol. The consultancy company handles relations with the client, including requirements analysis, data extraction and final project presentation, backed by our support. We then use Apromore to handle the data science on the backend, such as the extraction of analytics and the elaboration and interpretation of results.
After an initial feasibility study to raise internal awareness of the potential of Apromore and process mining in general, we identify internal sponsors who can be trained in the use of Apromore within the client organisation. This leads to capacity building for those organisations that intend to invest on running process mining initiatives on their own, on a systematic basis. In the case of Banca Intesa, they have put up a team of five business analysts that look after Apromore.
Secondly, we partner with organisations that support the development of Apromore’s functionality through hiring developers or contributing directly to Apromore’s codebase, including BusinessOptix (a London-based software company), Holocentric (a Sydney-based company integrating its business modelling tool with Apromore) and Leonardo (who are partnering with our team to refine the platform strategy and strengthen the codebase).
As the software is open source the codebase can benefit from contributions originating from different donors, collectively helping to improve Apromore’s quality and functionality and contributing to our pool of knowledge on process mining
The work we’re doing is more than just important industry-based partnerships. Beyond these structured initiatives, many academics and research students from around the world have contributed plug-ins or simply code fixes to Apromore. As the software is open source (under the LGPL 3.0 licence) the codebase can benefit from contributions originating from different donors, collectively helping to improve Apromore’s quality and functionality and contributing to our pool of knowledge on process mining.
Process mining is more than a tool; it is a solution tailored by its very nature to an organisation. Apromore allows organisations to focus on delivering value rather than on the means of process mining, by understanding how an organisation’s operations perform and how the latter can be effectively improved, one process at a time.