Comments for Markus Breitenbach http://blog.markus-breitenbach.com AI, Data Mining, Machine Learning and other things Thu, 11 Mar 2010 07:10:59 +0000 http://wordpress.org/?v=2.2.1 Comment on Validating patterns found by Data Mining techniques by Dan http://blog.markus-breitenbach.com/2007/02/22/validating-patterns-found-by-data-mining-techniques/#comment-26950 Dan Wed, 03 Mar 2010 17:24:23 +0000 http://blog.markus-breitenbach.com/2007/02/22/validating-patterns-found-by-data-mining-techniques/#comment-26950 http://arstechnica.com/science/news/2010/03/were-so-good-at-medical-studies-that-most-of-them-are-wrong.ars http://arstechnica.com/science/news/2010/03/were-so-good-at-medical-studies-that-most-of-them-are-wrong.ars

]]> Comment on Alternative measures to the AUC for rare-event prognostic models by Markus http://blog.markus-breitenbach.com/2010/02/16/alternative-measures-to-the-auc-for-rare-event-prognostic-models/#comment-26753 Markus Fri, 26 Feb 2010 17:40:09 +0000 http://blog.markus-breitenbach.com/2010/02/16/alternative-measures-to-the-auc-for-rare-event-prognostic-models/#comment-26753 The x-axis is just the score-value. The plot is supposed to show that the two classes overlap and can not be perfectly separated by the classifier. The x-axis is just the score-value. The plot is supposed to show that the two classes overlap and can not be perfectly separated by the classifier.

]]> Comment on Alternative measures to the AUC for rare-event prognostic models by steffen http://blog.markus-breitenbach.com/2010/02/16/alternative-measures-to-the-auc-for-rare-event-prognostic-models/#comment-26744 steffen Fri, 26 Feb 2010 10:47:44 +0000 http://blog.markus-breitenbach.com/2010/02/16/alternative-measures-to-the-auc-for-rare-event-prognostic-models/#comment-26744 Thanks for sharing this case study (including your thoughts). I have messed around with AUC and the Calibration-Refinement / Sharpness - measures also. A conclusion from various papers was, that Logistic Regression in general delivers well calibrated probabilities. Hence it was interesting to see the opposite can happen when only a small base rate is given. Side Note A: I did not understand the risk-score density plot. What is the x-axis ? Side Note B: http://home.comcast.net/~tom.fawcett/public_html/papers/ROC101.pdf is an excellent study of roc and auc in general. thanks again for this post. Blogs with technical and detailed data mining content are hard to find. Steffen Thanks for sharing this case study (including your thoughts). I have messed around with AUC and the Calibration-Refinement / Sharpness - measures also.

A conclusion from various papers was, that Logistic Regression in general delivers well calibrated probabilities. Hence it was interesting to see the opposite can happen when only a small base rate is given.

Side Note A: I did not understand the risk-score density plot. What is the x-axis ?

Side Note B: http://home.comcast.net/~tom.fawcett/public_html/papers/ROC101.pdf is an excellent study of roc and auc in general.

thanks again for this post. Blogs with technical and detailed data mining content are hard to find.

Steffen

]]> Comment on Strong profiling is not mathematically optimal for discovering rare malfeasors (on rare event detection) by Dan http://blog.markus-breitenbach.com/2010/01/10/strong-profiling-is-not-mathematically-optimal-for-discovering-rare-malfeasors-on-rare-event-detection/#comment-25420 Dan Fri, 22 Jan 2010 21:48:21 +0000 http://blog.markus-breitenbach.com/2010/01/10/strong-profiling-is-not-mathematically-optimal-for-discovering-rare-malfeasors-on-rare-event-detection/#comment-25420 The same paper is being discussed by the BBC: http://news.bbc.co.uk/2/hi/uk_news/magazine/8452260.stm The same paper is being discussed by the BBC: http://news.bbc.co.uk/2/hi/uk_news/magazine/8452260.stm

]]> Comment on Automation of Science by Dave http://blog.markus-breitenbach.com/2009/04/05/automation-of-science/#comment-25313 Dave Tue, 19 Jan 2010 21:38:43 +0000 http://blog.markus-breitenbach.com/2009/04/05/automation-of-science/#comment-25313 Older, but still pretty interesting: "A new theorem in particle physics enabled by machine discovery". http://dx.doi.org/10.1016/0004-3702(95)00128-X The note reports a novel finding in particle physics that was enabled by a machine discovery program. Older, but still pretty interesting: “A new theorem in particle physics enabled by machine discovery”.

http://dx.doi.org/10.1016/0004-3702(95)00128-X

The note reports a novel finding in particle physics that was enabled by a machine discovery program.

]]> Comment on Torpedo-Reviews in Machine Learning Conferences by Markus http://blog.markus-breitenbach.com/2009/03/01/torpedo-reviews-in-machine-learning-conferences/#comment-21825 Markus Mon, 21 Sep 2009 04:09:18 +0000 http://blog.markus-breitenbach.com/2009/03/01/torpedo-reviews-in-machine-learning-conferences/#comment-21825 This is hilarious: http://th.informatik.uni-mannheim.de/People/Lucks/reject.pdf Rejection letters for some of the most influential papers written in Computer Science. This is hilarious:
http://th.informatik.uni-mannheim.de/People/Lucks/reject.pdf
Rejection letters for some of the most influential papers written in Computer Science.

]]> Comment on Torpedo-Reviews in Machine Learning Conferences by Vitaly http://blog.markus-breitenbach.com/2009/03/01/torpedo-reviews-in-machine-learning-conferences/#comment-16568 Vitaly Fri, 10 Apr 2009 22:29:55 +0000 http://blog.markus-breitenbach.com/2009/03/01/torpedo-reviews-in-machine-learning-conferences/#comment-16568 Same thing happened to me just recently with my application to SIGIR. Since I trace my logs I could see that they even did not bother to try my tool, whereas I stressed that the most powerful part in my Demo presentation is to see how it really works. Same thing happened to me just recently with my application to SIGIR.
Since I trace my logs I could see that they even did not bother to try my tool, whereas I stressed that the most powerful part in my Demo presentation is to see how it really works.

]]> Comment on Can statistical models be intellectual property? by james http://blog.markus-breitenbach.com/2008/09/01/can-statistical-models-be-intellectual-property/#comment-14588 james Tue, 27 Jan 2009 11:13:47 +0000 http://blog.markus-breitenbach.com/2008/09/01/can-statistical-models-be-intellectual-property/#comment-14588 I saw http://blog.markus-breitenbach.com/2008/09/01/can-statistical-models-be-intellectual-property/ and wanted to mention a useful site: http://www.FreePatentsOnline.com It provides free patent searching, free PDF downloading, allows annoting documents and sharing them, and free alerts for new documents. If you have a spot, a link to let your users know abou the site would be great. I saw http://blog.markus-breitenbach.com/2008/09/01/can-statistical-models-be-intellectual-property/ and wanted to mention a useful site: http://www.FreePatentsOnline.com

It provides free patent searching, free PDF downloading, allows annoting documents and sharing them, and free alerts for new documents.

If you have a spot, a link to let your users know abou the site would be great.

]]> Comment on Computer Models and the Mortgage Crisis by Markus http://blog.markus-breitenbach.com/2008/09/28/computer-models-and-the-mortgage-crisis/#comment-13285 Markus Tue, 02 Dec 2008 17:06:29 +0000 http://blog.markus-breitenbach.com/2008/09/28/computer-models-and-the-mortgage-crisis/#comment-13285 <p>http://thedailywtf.com/Articles/The-Great-Excel-Spreadsheet.aspx</p> <p>In this case you didn't even have to lie in order to get bogus recommendations for the stock market...</p> http://thedailywtf.com/Articles/The-Great-Excel-Spreadsheet.aspx

In this case you didn’t even have to lie in order to get bogus recommendations for the stock market…

]]> Comment on CAPTCHAs - Not dead by Markus http://blog.markus-breitenbach.com/2008/08/14/captchas-not-dead/#comment-9967 Markus Thu, 21 Aug 2008 19:15:37 +0000 http://blog.markus-breitenbach.com/2008/08/14/captchas-not-dead/#comment-9967 Cool paper on segmenting the Microsoft captcha: http://homepages.cs.ncl.ac.uk/jeff.yan/msn_draft.pdf Cool paper on segmenting the Microsoft captcha:
http://homepages.cs.ncl.ac.uk/jeff.yan/msn_draft.pdf

]]>