Re: Distribution of results

On 11/1/06, Raghuraman K <raghuramank(at)aztecsoft(dot)com> wrote:
>
> Hi,
>
> I have a table like this: create table(xyz char(10), answer
> number(4)). There are a lot of rows in this table. I am looking at a query
> that will help me represent the distribution of data records based on the
> column answer. For example, we may take that the highest entry for answer
> column is 90 and the lowest is 2 and there are 1000 records. I am looking at
> a query that will tell how the 1000 records are distributed between the
> highest and lowest answer (in this case between 90 and 2). Can anyone
> please help?
>

It helps to know what kind of distribution information you are after.

Mean:
select sum(number) / count(*) from xyz;

Median:
Check out this URL
http://72.14.203.104/search?q=cache:kvZMBQuoAbkJ:people.planetpostgresql.org/greg/index.php%3F/categories/13-Math+postgresql+median+mean+functions&hl=en&gl=us&ct=clnk&cd=1&client=firefox-a

Range:
select max(number) - min(number) from xyz;

Population Variance:
select power(sum(number - mean), 2) / count(*)
from xyz
inner join (
select sum(number) / count(*) as mean from xyz
) as xyz_mean
;

Sample Variance:
select power(sum(number - mean), 2) / (count(*) - 1)
from xyz
inner join (
select sum(number) / count(*) as mean from xyz
) as xyz_mean
;

Note that I did not check the syntax for typos.

Anything more than this will require you whip out a Statistics book.

==================================================================
Aaron Bono
Aranya Software Technologies, Inc.
http://www.aranya.com
http://codeelixir.com
==================================================================