News

The previous KDnuggets poll asked the question:

Which data mining tools you regularly use?

Multiple choice of tools was allowed.  Based on voter logs, 495 people
voted for a total of 858 items, so an average data miner uses about 2 tools.
Admittedly, the results are unscientific, voters are self-selected, etc
(although there are multiple security measures to prevent people from voting
twice).  Nevertheless the results are interesting.

The most popular tools were:

        Clementine (156)
        SPSS/AnswerTree (135)
        SAS (104)
        CART/MARS (97)
        SAS EM (55)
        Megaputer (52)
        MATLAB (45)
        Angoss (29)
        IBM I-Miner (29)

(see full results and comments at
www.kdnuggets.com/polls/2001/dm_tools.htm)

It is interesting to compare these results to a similar poll
taken last year (see
www.kdnuggets.com/polls/2000/dm_tools_oct_2000.htm)

  Year		2000     Share	2001	Share	2001/2000
  ALL votes	 698		858		ratio

Clementine	120	17.2%	156	18.2%	1.06
SPSS/AnswerTree	105	15.0%	135	15.7%	1.05
  SAS		98	14.0%	104	12.1%	0.86
  CART/MARS	40	5.7%	97	11.3%	1.97
  SAS EM	81	11.6%	55	6.4%	0.55
  Megaputer	8	1.1%	52	6.1%	5.29
  MATLAB	--	--	45	5.2%	--
  ANGOSS	20	2.9%	29	3.4%	1.18
  IBM I-miner	51	7.3%	29	3.4%	0.46
  STATISTICA	14	2.0%	16	1.9%	0.93
  SGI Mineset	26	3.7%	14	1.6%	0.44
  Oracle Darwin	17	2.4%	14	1.6%	0.67
  Gainsmarts	32	4.6%	6	0.7%	0.15

About 20% more tools were selected this year, compared to last.

We see that the top tools: Clementine, SPSS, and SAS have about the same
share of usage.  Angoss slightly increased and Statistica slightly decreased
its share.

Among gainers we see:
      Megaputer, which increased its share 5+ fold,
      CART/MARS, whose share nearly doubled,
      MATLAB, which was not included in 2000 poll, but got a respectable
	      5% share in 2001.

Among losers are:
  Darwin lost about a 1/3 of its share.
  SAS Enterprise Miner, IBM Intelligent Miner, and SGI Mineset --
  all lost about half their share.
  Gainsmarts has lost 85% of its share.

One possible conclusion is that big and expensive server based
tools like SAS EM and IBM Intelligent Miner are being replaced by
cheaper PC-based tools like Megaputer, CART, and Angoss, whose
functionality could be sufficient for many tasks.

See full results and comments at
www.kdnuggets.com/polls/2001/dm_tools.htm