villebro commented on issue #27245:
URL: https://github.com/apache/superset/issues/27245#issuecomment-2058166386

   > @villebro That would probably work but I dispute the notion that #21093 is 
problematic. The issue in #21093 can be solved in the query with a IFNULL and 
the solution is disingenuous to the end user because it implies data is in a 
particular X value when the reality is no data is present. Consider a case of a 
sensor (for a car, weather, etc...) there is a big difference between the 
sensor returned no data and the sensor returned a zero value. The solution 
takes the position that 0 and null are the same.
   
   Actually I'm not so sure using an `IFNULL` would work here. The whole reason 
why these gaps appear is because the particular combination of grouping keys 
doesn't exist. Adding an `IFNULL(0)` would still leave a NULL value due to the 
pivot operation. Note, that there has been talk of doing away with the pivot 
operation for these charts, and that would likely make it easier to offer 
richer customization options. However, this requires very heavy redesigning of 
the charting system, and will likely not happen any time soon.
   
   > Real world example. In my day job one of the data sets my data engineering 
team works with is sensor data that comes off commercial airline flights 
looking for anomalies in mechanical parts. If we had implemented Superset we 
would have been in a world of hurt with this change because we need to track 
the cumulative sum of defects and also periods when the sensor is offline. The 
original view is correct and what we have implemented in our reporting platform 
(Tableau)
   
   While I understand your need for terminating the cumulative graph, I'm not 
convinced the common expectation is to see gaps in a cumulative graph where 
values are missing. I fired a quick "cumulative graph" in Google, and picked 
the first hit:
   
   
![image](https://github.com/apache/superset/assets/33317356/cc25cfd0-c530-4308-8101-3d7500a9c470)
   
   source: 
https://blogs.sas.com/content/iml/2020/03/25/read-cumulative-frequency-graph.html
   
   I feel this is a very typical cumulative distribution chart, and leaving a 
gap in it would IMO not be the expected result. Or maybe I'm misinterpreting 
what you're saying.
   
   Having said all this, I would still like to see if we can somehow achieve 
what you're after. Would you be able to share some redacted screenshots of the 
Tableau chart you mentioned so I can understand how the expected chart should 
look?


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to