Association rule in SQL Server Data Mining

Question

I have been working on a problem on association rules in SQL Server Data Tools (Visual Studio 2008) for quite a while but have not yet been able to figure out the solution.
The problem is: I have a table named Sales_history in my SQL database. This table has following columns: CustomerID, ItemID, Month (from May2012 to April 2013), QtyShipped. I am looking to find association between Items and i want to provide recommendation
to the Customer (In this case CustomerID) based on their purchases.
Note: there are around 630 customers and about 34000 products in my table. 
My approach:
I marked History_Table as both Key and nested. And in the Key, i checked CustomerID as input and Key whereas in nested, i checked ItemId as Key, input and Predict. 
When i run the model, i get a solution but i am not sure if i am configuring it right. Also i am not sure how i can write prediction join query to generate Item recommendations . I am really struggling with
this problem, eagerly waiting for the reply. Thank you.

Answer 1

Hi Tatyana,
Thank you so much for your reply. I have now been able to create the data mining model using association rule and by writing a DMX query, i am able to generate the item recommendations to be given to customers for items they have purchased. However, i have
noticed one thing that in the DMX query, it gives the same item recommendation for any item i put inside the query. 
Also, if i put any item in the DMX query from the generated list of recommended items, the output of that query also shows the item that is inputted inside the query.
Here is the query, that i am writing to generate item recommendations
SELECT predictassociation (CrossSellingModelV3.[Ztb Customer Item v3],INCLUDE_STATISTICS,5)
FROM CrossSellingModelV3
NATURAL PREDICTION JOIN 
(SELECT
(SELECT '17IS56126' as m )
AS Ms)
AS t
What can be the possible reason behind this? Is this something related to the kind of data i have? In my data, there are 632 distinct customers and 34000 distinct products. 
If i execute this query in management studio.
select customer_CD, COUNT(Item_CD) from ztb_Sales_History
group by customer_CD
order by 2 desc
the output shows that  there are some customers who have bought just 1 item and also there are customers who have bought 2400 items. i mean the range is very high.