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.

Metrics about the comparing log file and model

Hi, I need help to understand how the metric  Queued States and Num. States  are calculated. Could someone explain it?
this metric are availeble in ProM to export once using comparing model in petri net for example and log file.
I'm using plugin "by Replay a log on petri net for conformance Analysis.

Another metrics are about Only Reliable Results, im my opinion this should be easy to calculate but some points like value for Trace Fitness, Trace Lenght are no ok, or i did not understand how it calculate.

I appreciate any help

Rosângela


Comments

  • Hi Rosângela,

    Since the question is too specific, I think it's better that you contact the author of the plugin (the email is attached under the name of the plugin) to get a precise anwser.

    Guangming Li
    PhD student in Process Mining at Eindhoven University of Technology
  • hverbeekhverbeek Posts: 405
    Hi Rosângela,

    They're not really metrics, more like diagnostic information.

    To align a single trace to a model, a state space is required that combines both, and in that state space a shortest path is found. However, there is no reason to build the entire state space. The Num. States correponds to the number of states that were visited in the state space while looking for the shortest path, whereas the Queued States correspond to the number of states that are considered worthy of a visit (but have not been visited yet).

    For your other question I would need a specific example (model + trace) where you think the trace fitness is wrong. Basically, first the costs associated to the found shortest path are added. This is then divided by the total costs associated to a path that is known to be feasible: log moves for every event in the trace and the costs of aligning the empty trace in the model (which requires only model moves). This division is then subtracted from 1.

    Cheers,

    Eric.
Sign In or Register to comment.