EXPERT RESPONSE
There's really no perfect way to do this with just SQL.
The best solution is to use an extension to SQL
like Transact-SQL (Sybase and Microsoft SQL/Server) or PL/SQL (Oracle).
These language extensions allow you to write processing logic to cursor through
the table, fetching every 100th row using a loop counter.
This is not only the most accurate method but the most efficient as well.
If your table's memberid is an auto_increment or identity or sequence
column, and if it has relatively few gaps in the numeric sequence,
then you can come close to what you want with the following query --
select memberid, sex, age
from yourtable
where mod(memberid+27,100) = 0
Here mod() is the modulus function, which returns
a remainder upon division. Check your
database for the specific function name.
I added 27 to the memberid simply to introduce some randomness.
The query will select memberids 73, 173, 273, and so on, if they exist.
Out of a million rows, this should yield ten thousand,
fewer if there are gaps.
You see, a lot depends on what you want these
rows for. If you need to do a strict statistical calculation,
and need exactly every 100th row, you would want the first
solution. If you're just interested in getting approximately
one out of every hundred rows, more or less, for testing purposes,
then the modulus approach will usually suffice.
Finally, a recommendation I made in a previous answer,
Selecting random rows (29 May 2001):
The best solution, for situations where you only
need a small number of rows, such as in testing, is to have a
sample table which contains not random rows of live data, but
test cases -- in other words, a place where you can collect
those rows which are of interest or which represent conditions
you want to test for.
|
