sureli / naibx-mlc Goto Github PK

This is the companion library for the papers:

(short paper) Naive Bayes Classification for Subset Selection in a Multi-label Setting, Mossina and Rachelson, ICPRAI (2018)
http://oatao.univ-toulouse.fr/19557/
(long version) * Naive Bayes Classification for Subset Selection*, Mossina and Rachelson, arXiv preprint (2017)
https://arxiv.org/abs/1707.06142

The test/main.cpp file provides a script to run NaiBX.

As commented in the source, one must explicitly provide some input parameter, like the number of columns (e.g. dim_x=103) and the number of target labels available (e.g. labels=14). Please, for more information read the source code.

$ make naibx, will get you started

Data are assumed to be cleaned and ready for the analysis, that is, columns are either features (explicative variables) or labels.

The authors of mulan and MEKA very conveniently provide the community with a collection of multi-labeled datasets.

As an example, download the dataset yeast.arff from and run:

$ ./naibx data=path/to/yeast.arff dim_x=103 labels=14 all

Input parameters

• data=: path to your copy of yeast.arff
• dim_x=: number of features. This assumes
• all: computes all metrics available

$ ./naibx -bow data=/path/to/data.arff dim_x=103 labels=14

./naibx allows you to do k-fold cross-validation automatically, by providing the number of folds (e.g. kfold=10).