In the course of some of my research, I wanted to test training models using some of the ranking methods in RankLib. Because of the specifics of my data, optimizing a correlation coefficient is much more useful than the metrics currently available. Since I've had success with SVMrank, I figured that I'd try Kendall's tau (if I understand correctly, SVMrank basically can be thought of as optimizing this). I implemented this in RankLib as a new metric (see here: https://github.com/smsaladi/ranklib-kendall...
In the course of some of my research, I wanted to test training models using some of the ranking methods in RankLib. Because of the specifics of my data, optimizing a correlation coefficient is much more useful than the metrics currently available. Since I've had success with SVMrank, I figured that I'd try Kendall's tau (if I understand correctly, SVMrank basically can be thought of as optimizing this). I implemented this in RankLib as a new metric (see here: https://github.com/smsaladi/ranklib-kendall...
Move data for test files into tests directory. ...
code formatting mostly
Change all ''' to """ to make consistent throug...
additional files (comment same as previous commit)
many changes
namespacing is good!
remove temp file
move clean tmp folder script
Many changes to make this run successfully.
change randomint files to use NamedTemporaryFil...
ignore temp files
Remove c/cpp object files and ignore them in th...
remove unnecessary unafold files from the insta...
found RNA reference files in data folder.
Add data files for unfold. From github.com/peli...
namespacing is good!
ignore other sequence temp files
Move seeds into a list instead of many variable...
lots of reformatting. changing python file io f...
fix imports. Not sure why this was causing a pr...
move code in main to a test_* function and call...
rename files for pytest/unitesting compatibility
Move class diagram to tutorials folder
move tests to a folder recognizable by pytest
rename doc to tutorial
2to3 for all scripts
autopep8 -a for all files
Added v13 of tutorial document
Remove python cache files from version control ...
Thanks for your reply. Right, exactly. I am using a score for the relevance label....
Hello, Is is possible to use RankLib with more than 5 relevance labels in my data?...