SQL Server Clustering

[Shai Brumer]

Hi,

I have two questions:

1. I have a list of objects and a distance metric.
For example: object a and b have the distance of 0.7
(The data resides in a table with two id fields and a distance field)
I would like to create clusters of objects that are group by their distance.
Can I do it using SSAS (2005 or 2008)?

2. Is their a limitation on the number of mining columns being used for the
clustering algorithm?
Can it be hundreds\thousands\ millions?
On around millions of objects (1-10 million objects).

Thanks in advance,
Shai.

[Bogdan Crivat [MSFT]]

1. -- You cannot explicitly specify the distance for the SSAS clustering
algorithm. it determines the distance between points based on their
properties

2 It can be millions, typically by using a nested table (e.g. -- clustering
of customers based on the products they purchased, part of a million+
products catalog).
--
--
--
This posting is provided "AS IS" with no warranties, and confers no rights.
Please do not send email directly to this alias. It is for newsgroup
purposes only.

thanks,
bogdan

Hi,
1. I have a list of objects and a distance metric.
For example: object a and b have the distance of 0.7
(The data resides in a table with two id fields and a distance field)
I would like to create clusters of objects that are group by their distance.
Can I do it using SSAS (2005 or 2008)?
2. Is their a limitation on the number of mining columns being used for
the
clustering algorithm?
Can it be hundreds\thousands\ millions?
On around millions of objects (1-10 million objects).
Thanks in advance,
Shai.

[Shai Brumer]

First of all thanks for the answers.

Regarding question number two:
I would like to use the Microsoft Clustering to determine for me the number
of clusters to build (by setting the CLUSTER_COUNT to 0), would it be
possible to run the Algorithm with this setting on millions of mining columns?
Does this setting effect the duration of the run?
Does any setting effect the performance of the algorithm?

Thanks in advance,
Shai.

Post by Bogdan Crivat [MSFT]
1. -- You cannot explicitly specify the distance for the SSAS clustering
algorithm. it determines the distance between points based on their
properties
2 It can be millions, typically by using a nested table (e.g. -- clustering
of customers based on the products they purchased, part of a million+
products catalog).
--
--
--
This posting is provided "AS IS" with no warranties, and confers no rights.
Please do not send email directly to this alias. It is for newsgroup
purposes only.
thanks,
bogdan

Hi,
1. I have a list of objects and a distance metric.
For example: object a and b have the distance of 0.7
(The data resides in a table with two id fields and a distance field)
I would like to create clusters of objects that are group by their distance.
Can I do it using SSAS (2005 or 2008)?
2. Is their a limitation on the number of mining columns being used for
the
clustering algorithm?
Can it be hundreds\thousands\ millions?
On around millions of objects (1-10 million objects).
Thanks in advance,
Shai.