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.

Question about Conformance Checking / Heuristics Net

RobJanzenRobJanzen Posts: 11
edited June 2012 in Process Mining
Good morning,

I am having an interesting problem.  Using Prom6, I have a log file which I mined for a hueristics net and have a fitness of approx .7.  I have converted the heuristics net to a petri net and checked the fitness of the log used to create the heuristics net, and end up with 0.

Intuitively, this does not seem correct :-)

Can anyone shed some light on what I'm doing incorrectly, and give me suggestions as to how to proceed?  My goal is to generate a heuristics net, and to check other log files against it for conformance (checking to see how other data sets fit the process found from the original log).

Rob


Answers

  • aadriansaadrians Posts: 20
    Hi Rob,

    First, to my knowledge, fitness on heuristic nets is measured not in the same way as it is measured on petri nets. Petri net conformance checking in ProM 6 measure fitness based on optimal alignments between each trace in logs and model, while fitness on heuristic nets is not. The advantage of using optimal alignments is it is more robust to invisible tasks and duplicate tasks. This explains why you may have different fitness values.

    Second, fitness of 0 indicates that either the traces in your log totally not fit to your model, OR there is indeed no action is possible according to your model. I have the same issue when converting heuristic net in ProM 5 to petri net, in which the petri net has empty initial marking and therefore no action is possible according to the net. Check whether your Petri net has an initial marking. If you don't have any, use tools like YASPER (http://www.yasper.org/acpn/) to add an initial marking and load it back in ProM 6. Don't worry, PNML files that are exported from YASPER can be read by ProM 6, in the same way as when PNML files exported from ProM 5 can be read by ProM 6.

    Let me know how it goes :) 

    Arya Adriansyah
  • RobJanzenRobJanzen Posts: 11
    Thanks!  I'll try that and get back to you.
  • RobJanzenRobJanzen Posts: 11
    I'll have to do more playing around.  When I tried exporting the Petri Net as a .pnml, YASPER wouldn't read it.  (ProM would import it back in, so the file wasn't corrupt).

    I'm using a daily build, so I'll go back to the 'official' version to see if it works there as well as downloading the current daily build.
  • JBuijsJBuijs Posts: 912
    Hi Rob,

    As far as I know there is indeed an incompatibility problem when exporting a Petri Net from ProM 6 to Yasper.
    The other way round works indeed.

    As far as I know there is no way to get a Petri net from ProM to Yasper... Please let me/us know if you find one :)
    Joos Buijs

    Senior Data Scientist and process mining expert at APG (Dutch pension fund executor).
    Previously Assistant Professor in Process Mining at Eindhoven University of Technology
  • RobJanzenRobJanzen Posts: 11
    I reported the problem to Yasper, and they directed me to a new beta version http://www.yasper.org/beta/downloads/dotnet-2-0/Yasper-2.0b23.msi which is supposed to fix the problem.  It worked for me.  However, it is a beta, so your mileage may vary :-)

    Rob
  • JBuijsJBuijs Posts: 912
    Hi Rob,

    Thanks for sharing the (/a) solution! Will try it out for myself. Great to be able to transfer Petri net models between ProM and Yasper again.
    Joos Buijs

    Senior Data Scientist and process mining expert at APG (Dutch pension fund executor).
    Previously Assistant Professor in Process Mining at Eindhoven University of Technology
Sign In or Register to comment.