Rexer Analytics' 2009 Data Miner Survey

I'd like to alert readers to Rexer Analytics' 2009 Data Miner Survey. I urge you to participate by visiting the on-line survey at:


Rexer Analytics' 2009 Data Miner Survey

The Access Code is: TW4D2


Your entry is confidential and will help all of us better understand what is happening in the field of data mining.

Thanks!

Poll Results (Feb-20-2007): Toolbox Use

The Feb-20-2007 posting asked the poll question, Which Toolboxes from the MathWorks do you use (check all that apply)? After approximately 1 week, 38 votes were cast in total (including 1 by myself). The poll results follow:

1. Curve Fitting Toolbox (1 vote) 3%
2. Fuzzy Logic Toolbox (2 votes) 5%
3. Genetic Algorithm Toolbox (1 vote) 3%
4. Image Processing Toolbox (8 votes) 21%
5. Neural Network Toolbox (2 votes) 5%
6. Optimization Toolbox (6 votes) 16%
7. Signal Processing Toolbox (5 votes) 13%
8. Spline Toolbox (0 votes) 0%
9. Statistics Toolbox (13 votes) 34%

Not surprisingly, the Statistics Toolbox is the most popular among this crowd. The use of some of the other analytical toolboxes (such Neural Network and Fuzzy Logic Toolboxes) is also expected, though I was surprised both by the complete absence of anyone responding using the Spline Toolbox and the fairly good turn-out for the Optimization Toolbox. Given the popularity of the Jan-26-2007 posting, Pixel Classificiation Project, the 8 votes cast for the Image Processing Toolbox is expected.

Bootstrapping just the Statistics Toolbox results gives:


>> n = 38; p = 13 / 38; prctile(mean(double(rand(n,10000) < p)),[5 95])

ans =

0.2105 0.4737


So, with a 90% confidence interval, readers' use of the Statistics toolbox is somewhere between 21% and 47%.

Poll Results (Feb-04-2007): Toolbox Use

The Feb-04-2007 posting, Poll (Feb-04-2007): Toolbox Use featured the following poll question: Which MATLAB Toolboxes, if any, do you use? After 1 week, the poll has closed, with 27 Data Mining in MATLAB readers responding (1 of which was myself). The final results are:

1. None (base MATLAB product only) (0 Votes)
2. MATLAB Toolboxes from the MathWorks only (11 Votes)
3. MATLAB Toolboxes from other sources only (1 Votes)
4. MATLAB Toolboxes both from the MathWorks and other sources (15 Votes)

Interestingly, all reported using toolboxes. All voters but 1 indicated using toolboxes from the MathWorks.

How significant are these results? A quick bootstrap of 100,000 replicates yields the following:


>> Poll = [repmat(2,11,1); repmat(3,1,1); repmat(4,15,1)];
>> Replicates = Poll(ceil(27 * rand(27,1e5)));
>> Count2 = mean(Replicates == 2);
>> Count3 = mean(Replicates == 3);
>> Count4 = mean(Replicates == 4);
>> prctile(Count2,[5 95])

ans =

0.2593 0.5556

>> prctile(Count3,[5 95])

ans =

0 0.1111

>> prctile(Count4,[5 95])

ans =

0.4074 0.7037


So (roughly, due to limited precision provided by small sample size), with a 90% confidence interval, use of MathWorks Toolboxes only is somewhere between 26% and 56%, use of non-MathWorks Toolboxes only is between 0% and 11%, and use of both is between 41% and 70%.

Thanks to everyone who voted!

Poll (Feb-04-2007): Toolbox Use

I am curious about readers' use of Toolboxes (from the MathWorks or elsewhere) in their MATLAB programs. Please answer the following poll on Toolboxes:

This poll is closed.

See the poll results in the Feb-11-2007 posting, Poll Results (Feb-04-2007): Toolbox Use

Thanks for participating!