CJM-ab: Abstracting Customer Journey Maps Using Process Mining

Abstract

Customer journey mapping (CJM) is a popular technique used to increase a company's understanding of their customers. In its simplest form, a CJM shows the main customer paths. When dealing with complex customers' trajectories, these paths are difficult to apprehend, losing the benefit of using a CJM. We present a javascript-based tool that can leverage process mining models, namely process trees, and business owners' knowledge to semi-automatically build a CJM at different levels of granularity. We applied our approach with a dataset describing a complex process, and shows that our technique can abstract it in a meaningful way. By doing so, we contribute by showing how process mining and CJM can be put closer together.

Understanding complex process models by abstracting infrequent behavior

“…For each simplied model we have measured, using the 5 Using the plugin Matrix Filter in ProM with Mean as the Threshold adjusting Method. 6 Using the plugin Activity Filter: Indirect Entropy optimized with Greedy Search in ProM [14]. 7 We have also used other discovery algorithms such as Heuristics Miner [35] and ILP [36] among others, but the runtime and memory needed for the process discovery or to obtain the quality metrics make unfeasible to use these algorithms.…”

“…Also in the abstraction research eld, many techniques alter the granularity level of the data in order to abstract 3 the low-level activities of a log into high-level activities that are more understandable to the user [5,6,16,20,22,31]. These techniques, analyzed in [37], can be used to produce an abstracted event log, allowing to discover a model that describes a version of the process at a higher level.…”

Event abstraction in process mining: literature review and taxonomy

“…The technique subsequently finds these behavioral patterns in the fine-grained data, and, replaces them with a coarsegranular activity instance. In Bernard and Andritsos (2018), the author propose to use process trees, i.e., a subclass of Petri nets, to transform fine granular event sequences onto coarser-level activity instances. In particular, the authors propose to use event abstraction in the domain of customer journey mapping.…”

Mapping the Patient’s Journey in Healthcare through Process Mining

“…The authors proposed a customer journey exploration map using event logs and data analytics [22] as an opportunity to characterize hundreds or thousands of customer journeys at the same time. Bernard and Andritsos extended their work in [23], and developed a javascript-based tool that can handle process mining models to create a CJM at different levels of granularity; and additionally, they introduced the use of genetic algorithms to solve the problem of automatically building CJMs from event logs [24].…”