Data Mining / Analytic Software Tools

Poll

Data Mining (Analytic) tools you used in 2007: [534 voters] The first (narrow) bar corresponds to the number of votes where the tool was selected alone, and the second (wide) bar to the number votes where the tool was select as one among several; Tools are ordered in descending order of total number of votes Commercial Data Mining Software SPSS Clementine 116, 73 alone or with SPSS Salford CART/MARS/TreeNet/RF   106, 54 alone Excel 94, 2 alone SPSS 91, 49 alone or with Clementine SAS 80, 8 alone or with SAS E-Miner Angoss 78, 50 alone KXEN 70, 51 alone SQL Server 38, 2 alone MATLAB 30, 1 alone SAS E-Miner 25, 8 alone or with SAS Other commercial tools    21, 0 alone Statsoft Statistica 15, 2 alone Insightful Miner/S-Plus    14, 0 alone Oracle DM    12, 0 alone Tiberius 11, 3 alone FairIsaac Model Builder 3, 2 alone Xelopes 2, 2 alone Miner3D 2, 0 alone Bayesia 2, 0 alone Megaputer 1, 1 alone your own code 61, 7 alone Free Data Mining Software Yale 103, 70 alone Weka 48, 3 alone R    42, 0 alone Other free tools    30, 0 alone C4.5/C5.0/See5    14, 0 alone Orange    12, 0 alone KNIME    2, 0 alone

For comparison, here are the results of 2006 KDnuggets Poll on Data mining/analytic tools

Comments

data miner, Price versus value
I do not believe that bigger companies buy SAS due to quality, scalability and usability, but rather on inertia and the belief that if they pay the most money, they get the best product.
The value of a data mining workbench should be improved productivity through the use of innovative algorithsm that minimize the modeling effort.
Data Mining tools were never intended to be used only by statisticians and computer scientists. Unfortunately, most tools available in the market today require advanced skills in order to use effectively.
No - I don't want to dumb the user down... But simply adding more, equally inefficient algorithms into a data mining tool does not make it a better tool!
Value is based on improved productivity, not another novel algorithm.

(Editor: for fairness, free tools like Weka offer many more algorithms than commercial tools like SAS or SPSS. In my experience, companies buy commercial tools for many reasons, including better interfaces, reliability, ease of use, scalability - rarely just for more algorithms.)

yangdongzju
I start to use KXEN when i study datamining in zhejiang university. I think KXEN is very good at Algorithm compare to other tools such as WEKA,DB miner etc.

Terry Taerum, Cost / Value of information
I think the previous comment about the "Swiss army knife" is a good one. I suspect though the purchase of a data mining tool is driven as much by having the "right tool" as the cost of the tool. There is also the issue of "my language" and "my interface" and "my application" - which possibly explains why so many data mining applications come out of existing database and statistical programs (e.g. SQL, ORACLE, SPSS, SAS to name but a few...). Finally, there is the question of value of information and cost of information. Information tends to zero. That is, as it gets olders, as it gets known by others, as it disintegrates, it becomes less valuable. The cost also tends to zero. And this places a limit on what should be spent on an application. So in pulling a tool out of my favorite tool box, the right tool should be the most important issue from my point of view.

Ross Bettinger, Data Mining Tools
It would be interesting to cluster data mining tool chosen using size of company (revenue or equivalent measure) because the cost of the tool may be a factor in its adoption. SAS E-Miner is a Swiss army knife of capabilities but it costs a bundle, while a more moderately-priced tool that does not offer as many algorithms or features may be used because it is more affordable.