Data Mining Reference - CRISP - DM<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />I thought I'd offer this reference up to our SE team...I often get asked for a 'process model' for conducting data mining from my new to data mining customers. Linoff and Berry in "Data Mining Techniques" do an admirable job...but here's another one that I've shared with many during both my Kodak tenure and here with SAS. It's the CRISP - DM framework. In the mid to late 90's a group of people came together from industry, government and academia to try and formulate a structure and process model for data mining because there sentiment was:
a. It's the Wild Wild West out there with this new thing called 'Big Data' and b. It seems every vendor or statistical consulting hack has their own solution they are peddling.
So the group formulated this framework called Cross Industry Standard Process for Data Mining or CRISP-DM. It's vendor neutral, and statistical methodology agnostic. The group tried to come up with a 2.0 version a few years ago...but the trail appears to have gone cold. Nevertheless I think the original work still stands the test of time and is very useful for the novice or as a refresher for the more experienced. Give it a try...
Re: Data Mining Reference - CRISP - DMStrong ties to SPSS? Yes, in that they are publicly listed in the cover pages as a contributing author. But the document itself and the processes it describes are, IMO, vendor neutral. The group that put it together insisted on vendor neutrality wrt to the final product.
Re: Data Mining Reference - CRISP - DMDoesn't CRISP-DM have strong ties to SPSS? It was developed by Daimler Chrysler, SPSS and NCR, and SPSS references this throughout their documentation as the methodology for analytics.
ftp://ftp.software.ibm.com/software/analytics/spss/support/Modeler/Documentation/14/UserManual/CRISP-DM.pdf
