I was wondering if it is possible to somehow export an event log for a specific process from Camunda (Platform), i.e. I would like to export a log file (e.g., as CSV, even better a CSV conforming to the XES standard) for all executed instances (cases) referring to a specific process that were executed by Camunda. For a selected process, the events in the log should at least have the case id, activity/event names, timestamps of start and end, maybe the resource and other attributes.
The goal is to load this event log file into one of the many process mining tools out there (e.g., ProM, Disco, Apromore, you name it) and to analyse the process execution with the help of this tool.
There must be a way of exporting this type of event log already available in Camunda, right? Or do I need to write a specific adapter to talk to the history service somehow? I know of this work here: https://arxiv.org/pdf/2009.06209.pdf but it seems rather complicated for such a “common” functionality.
There are a bunch of ways of doing this - Camunda has an enterprise tool Optimize that in fact specializing in taking the data out of the engine and visualizing it, but you can also use the tool to export specific data to CSV.
Optimize does require the EE version or at least get a trial license so if you don’t want to do that, you can just use the REST API to get out the data you want and covert the response into something you can upload into a 3rd party system
Finally - you can just straight up query the Database directly and export the results, all the data you need is stored there - it just needs the right query.
Pretty bummed that you have not approached this application in the same way as with other camunda products (joint community/enterprise editions). Would be great to have a community edition that we could all contribute to…
I am desperately seeking a tool that can accurately “replay” process executions ontop of the bpmn model itself. This would be a great way to visualize flow on a case by case basis as opposed to the heatmap which is more appropriate for visualizing overall process activity.
Would be happy to take a stab at solving this for use in optimize if it were open source