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.

Problem using the Inductive Visual Miner with Deviations



Hello,

I am having some trouble understanding a result I get when I use the inductive visual miner.

The picture above pretty much shows my problem. In the extract of the graph produced by the inductive visual miner I have some deviations. The deviation not framed by the orange shape is a wanted result. But the Deviations which are framed by the orange shape confuse me. What exactly do they mean? Where do they come from? Is there a way to remove them?

Comments

  • sjjleemanssjjleemans Posts: 64
    edited February 2016
    Hi,

    The 'wanted' deviation shows that in 15 times, something was skipped while it should have happened according to the model. (a 'model move'). Each orange-highlighted deviation shows that in 2 times, something was executed while the model said it could not happen at that point (a 'log move').

    To find out what these deviations stand for, please click on the red line of the deviation. This will filter the event log and only keep the traces in which that deviation occurred. All numbers in the model will be updated. Moreover, the trace view (in the right panel, click 'traces') will also only show these traces.

    If you have any further questions, please do not hesitate to ask them.

    Kind regards,

    Sander Leemans
    Post edited by sjjleemans on
    Sander Leemans
    Assistant Processor (Lecturer) at Queensland University of Technology
    Author of the visual Miner and Inductive Miner

  • When i filter the results this is the log and the model I get. I am still having trouble understanding the log move. Does it mean in this case "B" and "D" are the log moves that cant be presented in the model and it is shown as deviation over "E" because "E" is the only event in these traces which can be presented in the model? What would happen if the traces contain different events (for Example "E" and "B") and only "D" would be the log move. Where would the deviation appear?

    Thanks for your help.
  • FDahmann said:

    Does it mean in this case "B" and "D" are the log moves that cant be presented in the model
    This is correct: B and D are the log moves. A clue to this is given by the line just outside the screen shot which says "only in log".

    FDahmann said:

    it is shown as deviation over "E" because "E" is the only event in these traces which can be presented in the model? What would happen if the traces contain different events (for Example "E" and "B") and only "D" would be the log move.
    Inductive visual Miner uses alignments to find an optimal match between a trace and the model. In general, log moves are projected as close as possible to their 'real' position, i.e. the place in the model where execution was when the log move happened.

    In this case, the alignment decided that the log move B happened during the execution of E. Specifically: the alignment chose the order to be E_start (invisible move), B_complete (log move), E_complete (synchronous move).

    I hope this answers your question.

    Sander Leemans
    Assistant Processor (Lecturer) at Queensland University of Technology
    Author of the visual Miner and Inductive Miner
  • anindita.bhowmikanindita.bhowmik Posts: 24
    edited April 2016

    Hi sjjleemans,

    ProM Version: ProM Lite

    I have been facing the same issue. I haven't been able to interpret the deviations in my log:

    I have attached a screenshot of the deviations in the petri net discovered on the inductive miner.

    The events recorded in my petri net are:

    1 level approval - PL ( Manual )

    1 level approval - PL ( Auto )

    2 nd level approval - MAC ( Manual )

    2 nd level approval - MAC ( Auto )

    3 rd level approval - BF ( Manual )

    3rd level approval - BF ( Auto )

    Request for clarification by any of the above approvers - Branch_FA_SEEK_CLAR 

    The deviations shown are:

     Capture_1. PNG and

    Capture_2.PG

    Requesting for your kind assistance in interpreting the "Only in model" and "Complete in log" deviations.

    Thanks and Regards,

    Anindita Bhowmik 




    Post edited by anindita.bhowmik on
  • JBuijsJBuijs Posts: 912
    Hi Anindita,

    The Inductive miner, after discovering the process model, replays the data on it. This results in three types of moves:
    1. a move which matches an event and an enabled activity in the model
    2. a move which is seen in the log, but which was not possible in the model. This is a 'move on log only'
    3. Vice versa, as you have shown in your screenshots, if a move is only in the model then an activity was expected but recorded in the event log at that location.

    Hope this helps.
    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
  •  Hi JBuijs,

    I thank you for the clear and simple explanation of my query. :)

    Here, I have an observation. The 'model only' moves which depict an activity which was expected to happen in the event log but didn't happen correspond to incomplete cases. I have recorded these traces with different activities named as "%%Approval_Stage%%_Pending".

    How do I filter out this traces based on these events which are "yet to be completed".

    There is a mechanism to filter specific traces once the model is created based on occurrence of an event in the trace. But,say I do not wish to include the traces with pending events in the model at all.

    What would be the approach for that ?

    Thanks and Regards,

    Anindita Bhowmik

  • Hi Anindita,

    Please apply the plug-in 'Filter log using simple heuristics' to filter out traces that do not end with specific end activities. On this filtered log, you can apply Inductive visual Miner again.

    Sander
    Sander Leemans
    Assistant Processor (Lecturer) at Queensland University of Technology
    Author of the visual Miner and Inductive Miner
  • anindita.bhowmikanindita.bhowmik Posts: 24
    edited April 2016

    Hi Sander,

    I thank you for the suggestions provided. I have an observation.

    Once the deviations are clicked and the traces are investigated the Claim request ID appears incomplete. Is this a known observation?

    Thanks and Regards,

    Anindita Bhowmik


    Post edited by anindita.bhowmik on
    Capture.PNG 37.7K
Sign In or Register to comment.