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.

Creating a petri net using ProM 6.8 for conformance testing against an event log

Hi, I am trying to create a petri net (see petrinet.docx attachment) from synthetic data so that I can use it as my normative model with conformance testing.
I am using the 'Mine Petri net with Inductive Miner' plugin. I have tried others but the results are not as good.
I can correctly generate a petri net for one knee using synthetic data (left.docx attachment). 
Though when I try to generate a petri net where surgery on the left and right knee can be done in parallel, this merges the left and right pathway after the arthroscopies, therefore losing the information of which knee has undergone an arthroscopy (both.docx attachment). 
The rules are: multiple arthroscopies can happen to the same knee. After a primary total knee replacement (TKR) a arthroscopy cannot be done on that same knee. A revision TKR and or an 'Attention to TKR' can be performed multiple times on a knee that has had a primary TKR.
I would be extremely grateful if anybody could let me know how to generate a petri net with synthetic data as the one in the first attachment.
Thank you! Sam

Answers

  • Hi Sam,

    The problem is that the only cases including both the left and the right knee contain only the arthroscopy activities (cases 5 and 6). As a result, the Inductive Miner can deduce that these two activities are indeed in parallel, but it cannot do that for the others. For example, there is no case where both a left and a right primary TKR are present. This results in the Inductive Miner deducing that there is a choice between them.

    Ideally, for every possible left-right pair of activities, there should be a case where one of them is followed by the other. And a left-right pair could be different activities, like a left arthroscopy and a right primary TKR.

    Kind regards,
    Eric.
  • Hi Eric,

    Thank you so much for your reply!

    I have now created a new event log as you suggest above. However, I have simplified it to only the arthroscopy and primary TKR events as everything past there is working.

    Still the left and right side merge after the arthroscopies.

    I have again attached my model and event log.

    Can you suggest anything else? Or maybe I didn't exactly follow your instructions, if so could you please advise me where I have gone wrong?

    Thank you

    Sam

      

  • Hi Sam,

    Clearly, your cases 15 and 16 cannot be reproduced by the model. They must be considered as noise by the Inductive Miner.

    Can you try to reduce the allowed noise level for the Inductive Miner? By default it allows for some noise. As your log does not contain noise, you could set this level to 0. If you start the Inductive Miner, it shows a configuration dialog with a slider for this level. Just slide it to 0 and then continue as usual.

    As an alternative, you could add cases containing only a  primary TKR followed by 'the other' arthroscopy. As far as I can see, a primary TKR is only followed by 'the other' arthroscopy in cases 15 and 16, which could be considered as noise by the Inductive Miner.

    Kind regards,
    Eric.
  • Hi Eric,

    Thank you again for your reply!

    For all my models I have always set the noise threshold to zero.

    I have taken your advise and also ensured that every combination of arthroscopy and primary TKR can be done in parallel and it has worked!

    This is only for the arthroscopies and primary TKRs, but once this is working I am sure the rest will be fine.

    Thank you so much, kind regards

    Sam

Sign In or Register to comment.