Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.programming > #16192

Re: Another little puzzle

From Tim Rentsch <tr.17687@z991.linuxsc.com>
Newsgroups comp.programming
Subject Re: Another little puzzle
Date 2022-12-30 05:07 -0800
Organization A noiseless patient Spider
Message-ID <86zgb5rqe6.fsf@linuxsc.com> (permalink)
References <puzzle-20221214131815@ram.dialup.fu-berlin.de> <algorithm-20221221130021@ram.dialup.fu-berlin.de> <average-20221228125821@ram.dialup.fu-berlin.de>

Show all headers | View raw


ram@zedat.fu-berlin.de (Stefan Ram) writes:

[edited for brevity]

> ram@zedat.fu-berlin.de (Stefan Ram) writes:
>
>> I waited a few days before answering to allow
>> sufficient time to think about the problem.
>
> I sometimes look at records of events like when someone got up
> or went to bed on different days and wonder about the "average
> time" he gets up or goes to bed.  Some years ago, I asked in
> another newsgroup how to average times, and from answers there
> I [found a long and somewhat forbidding formula]

> At that time, I did not understand what was going on!  In
> hindsight, today, I see that this seems to calculate the
> average via the average of the position of the two points on a
> 24-hour clock in two dimensions.
> 
> [..other references to the same general idea..]

Several people, including myself, thought of this idea, suitably
generalized to N points rather than two.  Two problems:

One, there is no way of knowing this idea is what you were
looking for.  Basically your puzzle was "Can you guess what I'm
thinking?", without giving any useful clues to solve the puzzle.
At the very least, it is misleading to call it a "puzzle".

Two, this approach gives unexpected (or counter-intuitive)
results for many ordinary cases where the times have at least
some clustering (for example, all times within a 4-hour range, or
potentially even a 15-hour range).  In cases like that, which are
likely to be common for most real-world data, using geometric
averaging gives an inferior result.

Besides those drawbacks, it looks like you didn't even bother to
read the responses.  As far as the outside world can tell, your
purpose in posting was just an exercise in self-congratulations.

I know you can do better.  And I hope that in the future you will
make an effort to do so.

Back to comp.programming | Previous | Next | Find similar


Thread

Re: Another little puzzle Tim Rentsch <tr.17687@z991.linuxsc.com> - 2022-12-30 05:07 -0800

csiph-web