I like to work with statistical information. But often people on the business site doesn’t really understand what i mean when i say the process has a standard Deviation of X-Days. Thats why i wanted to create a diagram which shows the Throuput time in days and if the process has a normal distribution.
After a lot of tryouts and fails i found the solution and i wanted to share it with you:
But what do i mean with the diagram? For a better understanding, i wanted something like this:
So i wanted a chartbased on my average throuput time and my standard deviation ( or sigma) to identify my outlayers in the process.
First i needed to design the dimension. For this i just put the workdays between function and the defined events for it:
Now i need to count the Cases for each of this dimensions as a KPI.
This KPI is defined as a linechart on the primary axis
than i needed to define the Median, AVG and Upper/lower STDEV for my chart on the secondary axis.
Because i wanted the overall Median and so on i used the global() function and made a round to the whole day. When its the same like the workdays_between of a specific set i just set the value = 1 else NULL.
CASE WHEN Global(ROUND(MEDIAN(WORKDAYS_BETWEEN(“CALENDER_ID”,“STARTEVENT”, “ENDEVENT”)))) = WORKDAYS_BETWEEN(“CALENDER_ID”,“STARTEVENT”, “ENDEVENT”) THEN 1 END
I did the same for the AVG and UPPER/LOWER STDEV(AVG + STDEV oder AVG - STDEV)
and my result looked like this:
Now i could see that my process is not even close to a normal distribution and that the stability and output of the process is massively influenced by the outlayers. As additional possibility i can set a filter on the outlayers and analyze them now.
I just wanted to share it here. Maybe someone has an idea what else could be optimized. I would love to get such a chart as own object in celonis or to add the avg,median and stdevs ( sigmas) in the through put calculator.
I found an issue in it. The global function ignores the selections in the analysis made. So when the user filters for example on processes which ran in the last year. The AVG and MEDIAN is calculated over all processes and not just the processes of the last year.