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.

Xes log data for industrial process

Hello all,

I want to generate *.xes log data out of a simulation to get a better insight into the simulated production process.
My production process consists out of jobs, tasks, and machines.
To log the data in *.xes format each job is a trace and each task in the job is an event.
Currently, I use the org  extension in *.xes and write the machine as a org:resource in each event.
For each event, I have a start and complete event (lifecycle:transition).
Additionally, I added a cost attribute from the cost extension for each event.
The goal in process discovery is now to find the machines with high costs and long processing times.
When I use the inductive visual miner and group by resource, I am able to find machines with long processing times.
However, I am not sure how I would be able to highlight the costs in this process.

The question now are:
  • How can I save the costs in the xes to highlight them in the process discovery?
  • Is there somewhere an overview of what ProM plugin uses which XML attributes (e.g. if I use the Id extension for events, how do I pass this information to the ProM plugins? 
  • What plugins are typically used to highlight specific attributes in the discovery phase?

Kind regards,
Alexander
 

Comments

  • Hello Alexander,

    The ProM plugins do not take all possible data attributes into account. Like the Inductive Visual Miner does nothing with the cost attribute you mention. Unfortunately, I am also not aware of any plugin that do take the cost attribute into account. Some plugins may be able to handle it like any other data attribute, but no plugin treats it like a proper cost attribute (again, as far as I know).

    • Including the Inductive Visual Miner, there are plugins that take the execution time of a task into account, and can show that. Although this is certainly far from ideal, it means that if you can 'model' the costs as the execution time of a task, these plugins could show these costs (be it as an execution time).
    • Unfortunately, the overview you mention does not exist. Some plugins indeed use some fixed set of 'known' data attributes, while other plugins may be more flexible and use a classifier instead. If using a classifier, the set of data attributes used depends on the log at hand.
    • Perhaps it would be possible to use data-aware discovery plugins. These plugins explicitly take data attributes into account, and may be more sensitive to, like, cost attributes.
    I hope this helps you a bit. Otherwise, of course, feel free to comment.

    Kind regards,
    Eric.
  • Dear Eric,

    thank you for your detailed answer.
    I will try the data-aware plugins and additionally I will try how much I can do with the classifiers.

    Kind regards,
    Alexander
Sign In or Register to comment.