Ideas for Grouping numeric data based on the distance from each other?

[Erik Nodland]

Hi Everyone,

Just after some ideas and suggestions on how to do this. I have a large
amount of numeric data which I would like to group given a distance as a
parameter.
IE. If my dataset was simply 1, 5, 9,10,11,16,17,25,29 and I pass in a
distance of 2 then I would want to end up with groups like
1, 5, (9,10,11), (16,17) ,25,29

So [9,10,11] is grouped because each number is within a distance of 2 from
each other. Equally [16,17] is grouped because of the same.

To make this more complicated each numeric would actually be a x, y plot. So
the actual data would be more like

[10,12], [20,21] , [21,23], [30,31] etc. Each one having 2 sets of numbers.
In this case the grouping would be

[10,12], ([20,21][21,23]), [30,31]. So [20,21] and [21,23] are grouped
because x in both sets are within 2 and y in both sets are within 2.

any body have any good ideas to start me off

Thanks very much in advance

SoS

[Peter Duniho]

In article <3C**********************************@microsoft.co m>
"ErikNodland" <er**@gotaabejoking.comwrote:

Just after some ideas and suggestions on how to do this. I have a
large amount of numeric data which I would like to group given a
distance as a parameter.
IE. If my dataset was simply 1, 5, 9,10,11,16,17,25,29 and I pass in
a distance of 2 then I would want to end up with groups like
1, 5, (9,10,11), (16,17) ,25,29

How should the algorithm group the data if you have a string of
numbers within 2 of an adjacent number, but where the entire range of
such adjacent numbers is greater than 2?

For example:

5, 7, 9, 10, 11

[...]
any body have any good ideas to start me off

It seems to me that as long as you can come up with a reliable,
unambiguous rule about how the grouping should work, it will be easy.
Otherwise, it will be impossible. :)

Pete

--
I'm trying a new usenet client for Mac, Nemo OS X.
You can download it at http://www.malcom-mac.com/nemo