Machine learning powered Process Mining
- Sponsored: Celonis SE
- Scientific Lead: M.Sc. Jan Philipp Thomsen, M.Sc. Juliane Sigl
- Project Lead: Dr. Ricardo Acevedo Cabra
- Semester: Winter 2017/2018
To avoid inefficiencies, bottlenecks and compliance issues, companies need to know what really happens in their business processes. Celonis, a Munich-based software company founded by three former TUM students, collects data of IT-supported business processes, reconstructs a visual map of the actual process flow and helps companies to spot problems affecting process performance. This data analytics technology is successfully applied by many global leaders, such as ABB, Adobe, Bayer, Siemens or Vodafone, to achieve greater operational efficiency. The next level of process mining integrates artificial intelligence and machine learning to automatically identify weaknesses in processes together with their root causes and prescriptive recommendations for how to improve efficiency even faster.
The project consists of four main components:
- Preprocessing: Understand common business processes (e.g., purchase-to-pay, accounts payable, IT service management)
- Extraction of event logs: Identify the relevant database structure, detect the right joins between tables, develop event logs and data models
- Design of new analyses: Develop new process mining analyses in Celonis, identify crucial weaknesses in industry processes, create business cases and define ways to improve processes
- Prediction of events: Create intelligent and predictive analyses to forecast which processes are most likely to cause problems and how to handle them accurately
Results
The results of this project were summarized in the final documentation and the final presentation