Model Selection in KNN - Githubissues

shogun-toolbox / shogun

Shōgun

http://shogun-toolbox.org

BSD 3-Clause "New" or "Revised" License

3.03k stars 1.04k forks source link

Model Selection in KNN #468

Closed iglesias closed 4 years ago

iglesias commented 12 years ago

Integrate in the model selection framework the K parameter (number of neighbors) of KNN. So far, it is recommended to use an specific method (classify_multiple_k) to choose the best k within a range of values.

For a shot discussion on the topic, please read te comments in

https://github.com/shogun-toolbox/shogun/pull/464

monicadragan commented 11 years ago

Hi,

Some questions on the issue:

What method do you recommend to use for model selection (any suggested paper)?
On what branch is developed "shogun/src/shogun/classifier/KNN.cpp" in Shogun? I couldn't clone it via git (only saw some old commits of the file).

Is anyone else working on this issue? I'd like to give a try.

Thanks, Monica

stale[bot] commented 4 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

stale[bot] commented 4 years ago

This issue is now being closed due to a lack of activity. Feel free to reopen it.