How to apply Data Mining (Association Rule) to a huge database?

What I want to do is to apply Association method of data mining on my SQL Server 2000 database. Association rule is something like "finding the most frequent items that appear together in database."

For those who don't know or who want to remember what is association method is like, take a look at this presentation about Association rule in Data Mining.

http://www.authorstream.com/Presentation/sushiltry-108428-association-rules-data-mining-science-technology-ppt-powerpoint/

17th slide gives a nice example of applying association rule on a database.

So Can you help me about how should I write my SQL c开发者_运维知识库odes (If that will be sufficient of course)

Thanks.

Edit: I have corrected the link.

You need to implement the apriori algorithm. Check link-1 and also link-2.

You need to use the Analysis Services of MS SQL Server. Analysis Services have the Microsoft Association algorithm which is an association rule mining algorithm. You need to provide the dataset, support and confidence value and the rest will be taken care by Analysis service. If the datset is very large, you may consider setting the pattern size.

http://technet.microsoft.com/en-us/library/ms174916.aspx http://technet.microsoft.com/en-us/library/ms167047.aspx

Alternatively, you can use your own implementation of association rule mining algorithms( Apriori, FP-Growth, Eclat, etc)

If you want some Java implementations of Apriori, FPGrowth, Eclat, etc., you could have a look at my website: http://www.philippe-fournier-viger.com/spmf/

But it does not offers SQL plugin, so to use this code, you would need to export your database in the appropriate format first.

I think that there may be some tools specialized for pattern mining in SQL databases. By searching you can probably find some.