To prevent spam users, you can only post on this forum after registration, which is by invitation. If you want to post on the forum, please send me a mail (h DOT m DOT w DOT verbeek AT tue DOT nl) and I'll send you an invitation in return for an account.

Multi-perspective Process Explorer - How to Find Wrong Events

Hi, Using Multi-perspective Process Explorer plugin, it show me some wrong Events in the Information Panel.
How can I find this wrong trace in log?
I apreciate any help

Thanks!


For example, im my analyse it show me 7 wrong events.

Comments

  • Hi, Nice to see that you are using my tool. 'Wrong events' means that there was an event in the event log (in some trace) that could not be matched to any of the transitions of the process model (so a move in log in alignment terms). These should show up as 'yellow' moves in the Trace View, which you can toggle with the button 'Toggle Trace View' in the DETAIL VIEWS box.

    From the picture I assume those events occur on the slightly yellow'ish coloured place at the larger flower structure. You can also click on that place to see that kind of events were missing, if I remember correctly.
  • Hi, Fmannhardt
    Thank you to yor answer, that is great i already see the indicate wrong events
    Your tool is the best plugin i ever use in ProM. thank you to put available for us a great tool.
    So i have a basic question, how did you measure the fitness ?

    I'm using a dataset with 17 traces, i would like to know how did the fitness is calculate for:
    In case 16: there is 1trace where the model did not matched. The fitness in this case is 83,33.
    in trace 17: there is 2 traces where the model did not matched. The fitness in this case is 66,67

    Thanks!
    :)






  • Sorry for the very late reply. I did not look into the forum for a while.
    The fitness calculation is explained here:
    http://dx.doi.org/10.1007/s00607-015-0441-1
    And probably easier to understand (I hope) in my just released thesis (Section 4.4, p92):
    http://repository.tue.nl/870754

    The basic idea is to take the cost of deviating moves (forced execution / wrong events or in alignment terms model move and log move) and normalize it by the worst case cost.

    If you take standard cost of 1 for every deviation, then the cost of deviating move is just the sum of deviating moves. Be aware that the MPE proposes a non-standard cost from what I remember. 

    To get the worst case cost for an alignment, we assume that every event in the trace is 'wrong', that is, appears as a log move (yellow) and that we take the shortest/cheapest path through the model with forced execution, that is, the cost of forced executions (model moves / purple) are added. The latter choice is somewhat arbitrary since we could take any path through the model. However, as we are interested in an optimal alignment the shortest/cheapest path would be preferred by such an alignment.

    Hope that this still helped.
Sign In or Register to comment.