Hi @florian.mrkvi,
Pull-up functions are meant to calculate the aggregation once and don't update based on (global) filter changes applied elsewhere in your analysis. I would like to believe that this never changed, but I wasn't around in Celonis around that time.
It would be strange if this was changed.
Kind regards,
Sverre Klein
Hi Florian,
I haven't experienced any behavior changes of PU_Functions since I am working with Celonis (which started in 2019 with CPM 4).
Only topic that comes to my mind was the change of how CASE WHEN behaves in February 2022:
Why are we planning this?
The current CASE WHEN behavior is not intuitive and contradicts the behavior of established SQL systems. The new behavior can simply be thought of as "NULL is treated as false". More information and details can be found here.
What will change?
When using an ELSE fallback condition, so far the rows evaluating to NULL would be returned as NULL even if the ELSE statement was correctly defined. Example, when <COLUMN> = ‘A’ evaluates to NULL, then the following statement
CASE
WHEN <COLUMN> = ‘A’ THEN ‘option A’
ELSE ‘other’
END
returns NULL instead of ‘other’. This counterintuitive behavior will change: CASE WHEN will return ‘other’ with NULL inputs. This will lead to less NULL results and more ELSE results if NULL values are input values, and if ELSE is being used. If no ELSE is defined or if ELSE NULL is used, the result will still be NULL and nothing will change.
BR
Dennis
Hi Florian,
I haven't experienced any behavior changes of PU_Functions since I am working with Celonis (which started in 2019 with CPM 4).
Only topic that comes to my mind was the change of how CASE WHEN behaves in February 2022:
Why are we planning this?
The current CASE WHEN behavior is not intuitive and contradicts the behavior of established SQL systems. The new behavior can simply be thought of as "NULL is treated as false". More information and details can be found here.
What will change?
When using an ELSE fallback condition, so far the rows evaluating to NULL would be returned as NULL even if the ELSE statement was correctly defined. Example, when <COLUMN> = ‘A’ evaluates to NULL, then the following statement
CASE
WHEN <COLUMN> = ‘A’ THEN ‘option A’
ELSE ‘other’
END
returns NULL instead of ‘other’. This counterintuitive behavior will change: CASE WHEN will return ‘other’ with NULL inputs. This will lead to less NULL results and more ELSE results if NULL values are input values, and if ELSE is being used. If no ELSE is defined or if ELSE NULL is used, the result will still be NULL and nothing will change.
BR
Dennis
Thank you Dennis!
I've also been working with Celonis since around this time and I remember the change in the CASE WHEN statement well! Recently I was thinking about the CASE WHEN crawler script which would have come handy to do something similar but Celonis already removed it from the ML-package - what a shame!
It's good to know that you can't think of a change in the PU functions. I was pretty sure about this but I still wanted to get a second opinion.
Thank you and have a nice weekend!
BR Florian
Hi @florian.mrkvi,
Pull-up functions are meant to calculate the aggregation once and don't update based on (global) filter changes applied elsewhere in your analysis. I would like to believe that this never changed, but I wasn't around in Celonis around that time.
It would be strange if this was changed.
Kind regards,
Sverre Klein
Yeah, but so much changed within the last couple of years. I think I got a little unsecure by the question :D
BR Florian