'AWS Quicksight - get std dev for fields grouped by sum?
New to QS. I have data like this:
| timestamp | count |
|---|---|
| 2022-01-01 | 5 |
| 2022-01-01 | 2 |
| 2022-01-02 | 0 |
| 2022-01-02 | 9 |
| 2022-01-02 | 2 |
I want to sum the counts for each timestamp, and then get the standard deviation for those sums.
My attempt was this, because a similar approach worked for getting the median:
stdev(sumOver({count}, [timestamp], PRE_AGG))
Resulting value is always a little lower than it should be(for example 13.53 instead of 15.03). What's wrong?
Solution 1:[1]
As suggested by @Register Sole it is probably the difference between the standard deviation of the sample, stdev in Quicksight which uses sqrt(N-1) and the standard deviation of the population, stdevp in Quicksight, which uses sqrt(N) as the denominator in the standard deviation formula.
Using your sample provided above I made the Quicksight Table visualisation below without any calculated fields, just adding the count field multiple times and changing the aggregation used for the column:
I confirmed the stdevs in both Excel and manually as accurate. Cheers (I would have added this as a comment but lack the rep)
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|---|
| Solution 1 | skabo |

