Page 326 -

308 12 Analyzing “Spaghetti Processes”

Fig. 12.7 Hierarchical clustering applied to heterogeneous event logs. The whole event log is
partitioned into smaller, more homogeneous, event logs. This process is repeated until it is possible
to create a “simple model” for each of the smaller logs. The resulting dendrogram can be cut closer
to the root or closer to the leaves. This reflects the trade-off between the simplicity of models and
the number of models

Fig. 12.8 Fuzzy mining applied to the event log of the housing agency. The cartography metaphor
is used to support seamless abstraction and generalization. Both models provide a view on the same
process. In the right model infrequent activities have been removed or amalgamated into cluster
nodes. Moreover, infrequent arcs are removed based on the selected threshold