Timeline for Using Percentile_Disc to compute Median/Interquartiles
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 10, 2015 at 21:30 | comment | added | Rob Farley | If you want an average of the two middle values, then DISC isn't want you want. Use CONT instead, but better would be to use a different method for performance reasons. | |
| Nov 10, 2015 at 20:40 | comment | added | Aaron Bertrand | @ijustlovemath I don't know how we can determine if 6.5 minutes is an acceptable runtime for your query. Is it acceptable to you? As Rob mentioned, many of us have blogged about median before; I ran some performance tests and determined that - at least in my case - no, the new SQL Server 2012 features were not the fastest. But it might not be true in your case, especially since I didn't test against Columnstore. You could test, though. | |
| Nov 10, 2015 at 20:08 | comment | added | ijustlovemath | Very good point! All of my values are distinct, and since I'm fairly simple in my needs, I think a simple AVG of the two middle elements would suit my needs. Is my query okay in that regard? Also, is 6.5 mins an acceptable runtime for such a query (columnstore indexed on 150M rows) | |
| Nov 10, 2015 at 20:05 | history | answered | Rob Farley | CC BY-SA 3.0 |