Dear Tobias,
Thank you for reaching out.
Please be informed that correlation is a lift correlation. For more information please check this link:
https://en.wikipedia.org/wiki/Lift_(data_mining)
Many thanks and best regards,
Tringa
Thanks for your Answer. I hope, that I didn't make a mistake. I try to understand the "correlation" with the P2P Practise Data-Set with 1,116,080 cases. It says that the "Umbrella Cooperation" has 50,716 violations and a correlation of 5.137 (see the attached screenshot).
To calculate the lift according to Wikipedia I calulate: lift correlation = (violating cases with "Umbrella Cooperation" / all cases with "Umbrella Cooperation")/(all cases with "Umbrella Cooperation"/all cases)
That means in practise: lift correlation = (50,716/72,448)/(72,448/1,116,080) = 10.784
Did I do a mistake while calculating?
Best regards
My understanding is the correlation expresses statistical significance. Have you had an opportunity to review help pages: https://docs.celonis.com/en/conformance-checker.html#UUID-c01ab242-ea87-4cf6-2423-2e0b523e98fb_id_ConformanceChecker-Performrootcauseanalysis
The described: "The percentage of conforming cases according to the defined process model."
The overview section provides statistics on overall process conformance, the numerical impact of non-conformance on KPIs, and deviations from the process model detected by the conformance algorithm.
My understanding is the correlation expresses the statistical relationship. Have you had an opportunity to review help pages: https://docs.celonis.com/en/conformance-checker.html#UUID-c01ab242-ea87-4cf6-2423-2e0b523e98fb_id_ConformanceChecker-Performrootcauseanalysis
Comment
lift = conformance % of all cases / conformance % of cases with umbrella corp; I believe we have a meeting set and look at this closer.
Dear Thomas,
the help pages did not provide an appropriate answer.
Thank you for trying to help me with my problem. I look forward to your response as soon as you are able to find the solution.
Best regards
I am sending an invite out - do you think we can connect for 1-minutes tomorrow?
Good newss! The math checks out
. I had 72228, and I needed to use 72448.
My fault I will close.
