Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations IamaSherpa on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

How to develop *very general* data mining algorithms *without* data?

Status
Not open for further replies.

whorl1quote1

IS-IT--Management
Feb 8, 2009
1
US
My company (a software company) is trying to build mining algorithms using/OEM commercial tools (like SPSS or SAS) to do some predictions for all our customers. the problem is that the we do not have any data (since our software is still in development) and the algorithm has to work for all diverse customers. Even when our product got implemented in the customer's enterprise and start collecting data, we allow them to model their business process in such dramatic different way.

the more typical process is that our customer should hire data analysts to develop prediction algorithms based on the massive data. The problem I am facing here is that we are trying to *automate* the tasks of data analysts and build a generic enough model that fits all customers.

For example, how can I determine a list of variables that will be having impact on the prediction? These variables are usually discovered as patterns when analysts start mining the data. These variables are also dependent on individual enterprise and the presented data.

I am so agonized over having to come up with a generic algorithm that'd fit all these situations. What I figure is that any complex algorithms won't make sense. The one make sense is probably as simple and universal as doing an AVG and take that value to predict the future.

Any comments are highly appreciated!
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top