News

Tej Anand and KDnuggets have run an experiment which offered a more detailed version of software feedback, including an ability to evaluate different software versions on ease of use and functionality.

However, unlike the previous software used by KDnuggets polls, the feedback agent software allowed users to vote multiple times which led to abuses. One of the interesting patterns discovered by Tej was that there was a large number of entries from one vendor, which ranked that vendor's software as excellent and competitive software as poor.

James Madison observed in Federalist papers (no. 10) in 1787 No man is allowed to be a judge in his own cause, because his interest would certainly bias his judgment, and, not improbably, corrupt his integrity.

He had a realistic view of human nature and would not have expected vendors to remain neutral in such a poll. However, I feel that the results of the feedback experiment have been compromised and do not represent a fair sample.

I still believe that valid polls of software use are useful and will look for better ways to conduct them. (Have suggestions? Please email to editor )

The other two questions in the feedback asked about the survey participants. It is not clear how objective those answers were, but majority of the responders were involved with data mining for more than 3 years and rated themselves either good or excellent data miners.

Here are the remaining questions:

How long have you been involved in data mining?

Answer	Number of responses
3 months	1
1 year	29
3 years	28
more than 3 years	81

Please rate your skills as a data mining practitioner

Answer	Number of responses
Excellent	41
Very Good	39
Good	34
Fair	20
Novice	4