[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: Adding an aggregate for variance

**From**: |
John Darrington |

**Subject**: |
Re: Adding an aggregate for variance |

**Date**: |
Wed, 14 Oct 2020 08:47:06 +0200 |

**User-agent**: |
Mutt/1.10.1 (2018-07-13) |

On Tue, Oct 13, 2020 at 09:47:18PM +0200, Jose E. Marchesi wrote:
Hi John.
Still, I see nothing wrong in having support for simple aggregates like
average, count, variance and the like. In their simplicitly, they can
be certainly useful for cases where using a full-fledged statistical
analysis tool would be overkill.
Fair enough. However average, count and variance are all just special
cases of what statisticians call "moments" [1], so instead of reimplementing
each one from scratch, why not save yourself several hundred lines of code
and simply implement a "moment" aggregate (with aliases if you like) which
will calculate any arbitrary moment? As a bonus, then you'll skewness and
kurtosis for free.
J'
[1] https://en.wikipedia.org/wiki/Moment_(mathematics)