Comments for TunedIT Data Mining Blog

Comment on There are no alternatives to data mining by Vladimir Nikulin on taking 2nd prize in Don’t Get Kicked | No Free Hunch

Fri, 03 Feb 2012 17:21:30 +0000

[...] In particular, some readers might be interested to consider text of my interview in Warsaw, Poland: http://blog.tunedit.org/2010/07/20/no-alternatives-to-data-mining/ This interview was given in June 2010. At that time, Kaggle was at the most early stages of [...]

Comment on Winners’ notes. Using Multi-Resolution Clustering for Music Genre Identification by jake

jake — Fri, 19 Aug 2011 23:03:23 +0000

Congratulations on your winning! I am learning as well! Thanks

Comment on Winner’s notes. Eleftherios Spyromitros – Xioufis on Music Instruments Recognition by music citizens

music citizens — Thu, 28 Jul 2011 01:02:03 +0000

Does your study tests music gifted individuals who are superb in this field?

Comment on Winners’ notes. Using Multi-Resolution Clustering for Music Genre Identification by Freddy López

Freddy López — Mon, 06 Jun 2011 21:18:49 +0000

Congratulations, Amanda Schierz, Marcin Budka and Edward Apeh. It has been fantastic read your strategy and learn… we are always learning

Comment on Winners’ notes. Brian Jones on Incremental Transductive Ridge Regression by Rohan Anil

Rohan Anil — Sun, 10 Apr 2011 03:06:26 +0000

Excellent Job! and Thanks for sharing.

Comment on Ed Ramsden on his winning solution in SIAM SDM’11 Contest by Ed Ramsden

Ed Ramsden — Mon, 07 Mar 2011 07:40:41 +0000

The final ‘training’ Youden score for my submitted model was 0.794, while the test score was 0.689, so there was a pretty significant difference. Because of the ensembling I did, the final training score came about as a result of the combination of sub-models that were trained to maximize their individual training Youden scores. The plot I put in my post gives a pretty good idea of the spreads between ‘train’ and ‘final test’ Youden scores for submodels of different dimension. I was only able to do this kind of post-analysis after the final result file was made public (finalDecisions.csv) and I knew what the real answers were.
ER

Comment on Ed Ramsden on his winning solution in SIAM SDM’11 Contest by Nicole

Nicole — Thu, 03 Mar 2011 19:11:04 +0000

Guess I should look at the post first. Then I can see my approach is viable, maybe I can use it in another competition.

Comment on Ed Ramsden on his winning solution in SIAM SDM’11 Contest by Nicole

Nicole — Thu, 03 Mar 2011 18:00:40 +0000

I wonder if Ed would share the balanced Youden index he was achieving with the training data. I did not have time to make an entry, but have developed my own approach as practice. I do not know how to judge my model. Did you find a large or small difference between the training result and the final result? Thanks for any info you can share. I know I am not overfitting my data. I am using a Bayesian approach with only a couple of ‘knobs’ to turn.

Comment on Winners’ notes. Frank Lemke on self-organizing high-dimensional QSAR modeling from noisy data by Tweets that mention Winners’ notes. Frank Lemke on self-organizing high-dimensional QSAR modeling from noisy data « TunedIT Data Mining Blog -- Topsy.com

Fri, 18 Feb 2011 00:23:22 +0000

[...] This post was mentioned on Twitter by Sooyoung Oh, TunedIT. TunedIT said: Winners' notes. Frank Lemke on self-organizing high-dimensional QSAR modeling from noisy data: http://t.co/MG3opD8 [...]

Comment on About by Marcin Wojnarski

Marcin Wojnarski — Tue, 14 Dec 2010 11:36:09 +0000

Hi Alex, we’re very glad you like TunedIT Blog. What do you mean by interchange of news and information? This sounds interesting, but can you say something more?