Resource and activity clustering based on a hierarchical cell formation algorithm

Landelin Delcoucq, Thomas Dupiereux-Fettweis, Fabian Lecron, Philippe Fortemps

Abstract: In this paper, we focus on the resource perspective of process mining and more precisely on the clustering of resources sharing the same behaviors. This problematic was addressed through the use of a well-known facility layout method: cell formation. We propose an algorithm combining the resource perspective and cell formation approach to make the best use of their respective features. We wish to identify both subgroups of resources that perform similar activities and subgroups of activities performed by common resources. This new hierarchical approach provides new insights into the clustering problematic because of its bi-dimensional clustering. Experiments are considered on synthetic and real data.