Process Mining best practices

Hello,
We want to implement Process Mining for an end-to-end process that includes around 15 applications. Out of these, there is an application without reliable timestamp information. Has anyone ran into situations like these? If so, how did you handle it.
Also are there any best practices to implement UiPath Process Mining. Please share if you have any details on these.

Thanks

hi @fundoovj
I am solving a similar scenario when I have two different activity logs (from 2 different applications) that together create one process. How do you go about connecting these two logs so that they reliably represent the process? Only by putting together the timestamp?
Thanks!

The only piece of advice one could give here is to take a step back from tech and try to understand the business process. When my process was based on different applications, I was able to reproduce and validate time stamps by making assumptions. E.g. when application 4 has bad or missing time stamps and application 3 and 5 are fine, you could try and assume based on typical process step duration and information from better documented steps before and after, what the missing time stamp would be. Classic interpolation if you will.

I would suggest googling your way through actual ETL/Data Integration/Data Management forums, as this is a typical challenge in these kinds of projects and has been solved in lots of different ways.

Also, what I have learned is, in the case of transaction data, you might always be lucky and find another source (hidden table somewhere deep in technical db schemas), so take a good look at target applications documentation - maybe reach out to their tech support…