Is there any plan to add new techniques to improve the data mining part of SQL Server

I have worked with many data mining softwares and found SQL Data mining part of Microsoft product needs to imrove a lot to whoo the customers.

We continue to advance the data mining functionality in future versions of SQL Server. What aspects do you feel need improvement in order to "whoo" customers?

Thx

-Jamie

|||I think It'd be great if you guys were able to implement a little more in the way of the IDE telling you what was/wasn't possible... Such as not being able to create a lift chart for association models, and describing cases/nested tables for example. Or simply refer people to the MSDN documentation. Also, improving the UI, at least for the predictions tab would be nice. One thing I've noticed that annoys me is that when you hit the results button when writing DMX from the "SQL Page", it switches the button to automatically go to the design view, rather than the SQL page again. So everytime I must click on the small dropdown arrow and click "SQL" again...and again. Overall it's very functional and seems to be a big step up from the last version!|||

I will explain.

1.If you add more algorithms like logistic regressions or regression models, Machine learning (both supervised and unsupervised learnings), more advance decision trees algorithms(like CHAID, CART, QUEST and EXHAUSTIVE QUEST), more associative modelings like market basket analysis & sequential modeling, pricipal componenet analysis, it will be great and all algorithms are very good kind of decision making algorithms in respetive fields like Finance, Banking, market research, retail, telecom, CRM etc. For eg: If take risk modeling in Banking or finance domian, the logistic regression model plays an important role to score the default risk.

2. I have seen that all major players in Data mining like SAS or SPSS, they have features to connect with MS SQL SEVRVER and share the modeling features of later. Is there any feature vice versa?

3. Data mining plays a key role in modern CRM. If you have a better data mining features and you can have better CRM analytic software itself.

4. Text mining plays a key role in call center data analysis, Credit card analysis, telecom data modeling and even we can use this for better web mining. Why can not you try this kind of advanced features.

I think this explationation sufficient for your question to whoo customers. A better decision making algorithms and its proper implimentation in time and monitoring helps companies to save millions of dollors, in single year itself (eg: citi corp, DSP Merril lynch, Chase, Wal-Mart, Target (Even I was in the analytical team for some time), etc).

Thanks

Ajesh

|||

Thanks Ajesh - as Jamie says, we will continue to invest in improving the features of SQL Server data mining. It is great to get feedback from people on the forum or in mail that helps us to plan future versions. And, in fact, some of what you are asking for is already in the product - logistic regression, association rules (for market basket analysis) for example. Sharing of models is enabled between different tools mostly by PMML which SLQ Server, SPSS and SAS all support in our own ways.

Data mining does indeed play an important role in CRM. Microsoft Dynamics CRM has released a special "Analytics Foundation" which enables CRM users to integrate SQL Server Analysis Services (OLAP and Data Mining)

You can read more about the Analytics Foundation here:

http://www.microsoft.com/dynamics/crm/product/analyticsfoundation.mspx

And there is an interesting article here in destinationCRM about the role of data mining in the future of CRM and Microsoft's impact: http://www.destinationcrm.com/articles/default.asp?ArticleID=6833

Text mining is available in SQL Server through the SSIS text mining components. See the tutorial here: http://www.sqlserverdatamining.com/DMCommunity/Tutorials/default.aspx

So, as you can see, we have made some very significant investments in these areas, and we fully expect to continue that momentum through many releases to come.

|||One can also add features to support Ontologies, that can be useful for semantic analysis. Also a standardised ontology representation could be maintained to enable any kind of APIs to access the database as and when required.|||

Hi

I really agree with Jamie and Donald but I have a suggession that if you can integrate data mining, text mining and webmining into one platform or one single module, this will be helpful for thousands of users.

Thanks

Visiting lecturer

Madras University-Chennai

India

Also Analyst

JDA Software India, India

|||

Thanks - that is certainly something we should consider for the future.

When you say "web mining" are there specific features and functions that you would like to see?

|||

Hi

I am suggesting the following 2 concepts where you can look into.

1. web mining for business analysis
2. web mining for technical analysis

Also I appreaciate special functional enhancements for applications like

1. Customer profiling
2. personalization
3. market segmentation
4. Target marketing
5. Cross-selling
6. Integration with CRM

Over all a "COMPLETE E-BUSINESS MARKETING SOLUTION" and "Implementaion methodology for solutions". This kind of solution already exist for DataStage, etc.

Thanks

Ajesh