New data sources for analytics

My column scheduled to appear July 11 (available via this link to my column archive starting at some point that day) points out the importance of new kinds of data, and in particular new kinds of data that are explicitly created to be used in analytics.  

Here are some overflow thoughts on that subject.

1.  Most of the examples I gave were in the area of CRM and/or product quality.  This fits; most commercial applications of statistical analysis are in those areas.  Quality came first, but CRM is the larger application area now. 

This is a recurring theme; CRM and quality also run #1 and #2 in the specific market for text mining.

2.  However, the example that actually triggered the idea to write this column was medical research across a database of patient records.  Gary Kennedy of RemedyMD led me to the idea.  And last Saturday's Boston Globe wrote of a $20 million Harvard project to do this on an enormous scale.

3.  One of my main suggestions was to actively solicit extra customer feedback.  This is a big part of the SPSS story, in what they call "market research" but might better be named something like "opinion surveys".

4.  I mentioned explicit testing to help measure price elasticity.  Zilliant has that baked into it's pricing/revenue management product line. (More on Zilliant and Vendavo another time.)

5.  The Globe has actually had two highly relevant front page stories in a three day period.  The other is today's revelation of a program to gather data at every traffic stop that be helpful in determining whether racial profiling is occuring.  This is a great idea, although Ironically I wonder whether text mining of existing data might not already give them a big clue.