[ARROW-10070] [C++][Compute] Implement stdev aggregate kernel

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.0.0
• Component/s: C++
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/26089
• Assignee: Yibo Cai
• Reporter: Yibo Cai
• Created: 23/Sep/20 04:15
• Updated: 11/Jan/23 08:10
• Resolved: 30/Sep/20 12:24

Description

To calculate standard deviation of an array or chunked array.

I would prefer two pass algorithm [1] as a balance of numerical stability and performance. Numpy uses this method to calculate variance [2].
Welford's online algorithm [3] is more stable, but also more expensive in computation.

[1] https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Two-pass_algorithm
[2] https://github.com/numpy/numpy/blob/92ebe1e9a6aeb47a881a1226b08218175776f9ea/numpy/core/_methods.py#L176
[3] https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Welford's_online_algorithm

Issue Links

• is duplicated by: ARROW-3122 [C++] Incremental Variance, Standard Deviation aggregators
• links to: GitHub Pull Request #8269