Benchmarking Classifier Performance with Sparse Measurements
Jan Motl · ITAT · 2015
WWW home page: http://relational.cvut.cz Abstract: The presented paper describes a methodology, how to perform benchmarking, when classifier perfor- mance measurements are sparse. The described methodol- ogy is based on missing value imputation and was demon- strated to work, even when 80% of measurements are missing, for example because of unavailable algorithm im- plementations or unavailable datasets. The methodology was then applied on 29 relational classifiers & proposi- tional tools and 15 datasets, making it the biggest meta- analysis in relational classification up to date.