Sequence clustering

[keropi]

Hi, I've just begin learning data mining in SQL server 2005 and trying
to predict the sequence in which customers will take up products. but
the problem is that from the prediction, i get for example product A
as sequence one, then from sequence 2 to 10, i get all of the same
product. but i can see from my data that there are no customers that
are like that.

Can someone guide me to see where i went wrong?

Thanks so much

[Dejan Sarka]

Post by keropi
Hi, I've just begin learning data mining in SQL server 2005 and trying
to predict the sequence in which customers will take up products. but
the problem is that from the prediction, i get for example product A
as sequence one, then from sequence 2 to 10, i get all of the same
product. but i can see from my data that there are no customers that
are like that.
Can someone guide me to see where i went wrong?

I guess you did not define key, input and predictable columns correctly.
Let's say, for example, you have Orders and OrdeDetails table. define Orders
as case and OrderDetails as nested table; OrderId fro Orders and ItemId from
OrderDetails as key, ItemId as input as well, and Product as input and
predictable.

[keropi]

Hi,
I think it was a key issue.

Lemme try again.

Thank you.

Post by Dejan Sarka

Post by keropi
Hi, I've just begin learning data mining in SQL server 2005 and trying
to predict the sequence in which customers will take up products. but
the problem is that from the prediction, i get for example product A
as sequence one, then from sequence 2 to 10, i get all of the same
product. but i can see from my data that there are no customers that
are like that.
Can someone guide me to see where i went wrong?

I guess you did not define key, input and predictable columns correctly.
Let's say, for example, you have Orders and OrdeDetails table. define Orders
as case and OrderDetails as nested table; OrderId fro Orders and ItemId from
OrderDetails as key, ItemId as input as well, and Product as input and
predictable.
--
Dejan Sarka
http://blogs.solidq.com/EN/dsarka/default.aspx