[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: The Prize #3
When grilled further on (Sun, 19 Sep 2004 16:46:25 +0930),
Fraser Farrell <fraser@trilobytes.com.au> confessed:
> tom wrote:
> > 6) The improvement must be real. We Will figure out a way to test the
> > result. In other words, no hidden code that makes the result look good.
>
>
> One useful check of the numbers would be a test for conformance with Benford's
> Law. Which nowadays is widely used to check for fraud involving massive
> collections of numbers. For an introduction, read these:
Hmmm... In all the queries below, fd column is 'first digit'.
I didn't expect the magnitude to work, but the error worked nicely. Of course,
I suspect I could fake these now that I know about this ;-)
My magnitude range for V covers 2.84 to 16.3791, I covers 1.89 to 14.60
tasv=# select count(*), substring( mag_v::text from 1 for 1) as fd from
catalog where mag_v != 99 group by fd order by fd;
count | fd
--------+----
691920 | 1
1 | 2
5 | 3
39 | 4
159 | 5
1039 | 6
4213 | 7
14633 | 8
40641 | 9
tassiv=# select count(*), substring( mag_i::text from 1 for 1) as fd from
catalog where mag_i != 99 group by fd order by fd;
count | fd
--------+----
607613 | 1
9 | 2
43 | 3
164 | 4
786 | 5
3618 | 6
15105 | 7
38370 | 8
86765 | 9
Moving on to the smag (error from DAOPHOT):
tassiv=# select count(*), substring( smag::text from '[1-9]' ) as fd from obs_v
group by fd order by fd;
count | fd
---------+----
2064216 | 1
1663100 | 2
1444893 | 3
1165236 | 4
928826 | 5
742990 | 6
592820 | 7
468684 | 8
373451 | 9
tassiv=# select count(*), substring( smag::text from '[1-9]' ) as fd from obs_i
group by fd order by fd;
count | fd
---------+----
1971130 | 1
1651006 | 2
1524232 | 3
1265356 | 4
980525 | 5
737823 | 6
553203 | 7
424793 | 8
336634 | 9
--
09:13:33 up 13 days, 21:17, 4 users, load average: 2.04, 2.15, 2.16
Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004
PGP signature