Open bittlingmayer opened 8 years ago
As far as I know, there is no such list. In QuEst, you usually perform feature selection during model training :)
Gustavo Henrique Paetzold Ph.D. Candidate in Computer Science University of Sheffield
De: Adam Mathias Bittlingmayer notifications@github.com Enviado: sábado, 1 de outubro de 2016 08:14:55 Para: ghpaetzold/questplusplus Assunto: [ghpaetzold/questplusplus] Importance of features? (#36)
Is there any list of the features ranked or reduced by importance? Or are those features that are in the source precisely those which were determined to be necessary for optimal performance?
We would like to submit and compare our own ranking. If it varied significantly for the same training data, that could mean one of us has a bug or other subtle difference in implementation.
You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/ghpaetzold/questplusplus/issues/36, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AHrSdzhrFVguFafYRGKmrcSa37D-atPxks5qvhZ_gaJpZM4KLs4Q.
Hi,
We've done lots of feature analyses (see e.g. http://link.springer.com/article/10.1007%2Fs10590-014-9164-x) , but the ideal set is often dependent on your data/labels, so what Gustavo is suggesting is the best option.
Best, Lucia
On 1 October 2016 at 10:29, Gustavo Henrique Paetzold < notifications@github.com> wrote:
As far as I know, there is no such list. In QuEst, you usually perform feature selection during model training :)
Gustavo Henrique Paetzold Ph.D. Candidate in Computer Science University of Sheffield
De: Adam Mathias Bittlingmayer notifications@github.com Enviado: sábado, 1 de outubro de 2016 08:14:55 Para: ghpaetzold/questplusplus Assunto: [ghpaetzold/questplusplus] Importance of features? (#36)
Is there any list of the features ranked or reduced by importance? Or are those features that are in the source precisely those which were determined to be necessary for optimal performance?
We would like to submit and compare our own ranking. If it varied significantly for the same training data, that could mean one of us has a bug or other subtle difference in implementation.
You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/ ghpaetzold/questplusplus/issues/36, or mute the thread< https://github.com/notifications/unsubscribe-auth/ AHrSdzhrFVguFafYRGKmrcSa37D-atPxks5qvhZ_gaJpZM4KLs4Q>.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ghpaetzold/questplusplus/issues/36#issuecomment-250902896, or mute the thread https://github.com/notifications/unsubscribe-auth/ABRaBDGIROTjDMb0WDd3fClCS8B-ECxxks5qvifjgaJpZM4KLs4Q .
Lucia www.dcs.shef.ac.uk/~lucia/
Is there any list of the features ranked or reduced by importance? Or are those features that are in the source precisely those which were determined to be necessary for optimal performance?
We would like to submit and compare our own ranking. If it varied significantly for the same training data, that could mean one of us has a bug or other subtle difference in implementation.