Provide additional variant classifications besides the “Unique” classification that currently is used. To summarize:
1)Variant class 1 (Existence + Frequency + Sequence) (current method)
2)Variant class 2 (Existence + Frequency)
3)Variant class 3 (Existence)
In most productive process mining projects end-users will typically see thousands of different variants. A frequent complaint is that the current variant list and numbers give a false impression on the process stability. Its considered too strict…
The other options would allow for more ‘meaningful’ variants.
Of course ‘meaningful’ is a subjective term but I'd prefer additional views on the process variance where I can select what best fits to the current analysis.
One further comment.
Additionally i can also envision an option that we offer a variant customization tab in our Data Set configuration - similar to the Conformance tab that is now also available.
Before loading the data the user can configure a custom variant classification. Here the user can do 3 things
1- Select the functions that need to be considered in the variant determination
2- Select the variant classification (see first comment) that needs be used
3- Give custom variant class a name
After which the user then can load the data, the system determines the variants, In analysis the user can select his/her own variant classification to analyse the "meaningful" variants.