Has anyone used RankLib for classification? I was wondering what is the best way to structure your input file. Use target values as 1 and 2 for class 1 and 2 respectively.
say we have one labeled point with label with class label 1 we generate two feature points in the feature file as shown below
e.g.,
1 f1:v1 f1:v2 # Editorial label of 1
2 f1:v1 f1:v2
Is this the right approach?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
No, it is not :-) If you want to do classification, use tools that are designed for classification. RankLib is for ranking, which uses a different set of optimization criteria. It won't give you good classification performance.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Has anyone used RankLib for classification? I was wondering what is the best way to structure your input file. Use target values as 1 and 2 for class 1 and 2 respectively.
say we have one labeled point with label with class label 1 we generate two feature points in the feature file as shown below
e.g.,
1 f1:v1 f1:v2 # Editorial label of 1
2 f1:v1 f1:v2
Is this the right approach?
No, it is not :-) If you want to do classification, use tools that are designed for classification. RankLib is for ranking, which uses a different set of optimization criteria. It won't give you good classification performance.