1. Towards Better Evaluation of Multi-Target Regression Models
- Author
-
Hendrik Blockeel, Evgeniya Korneva, Koprinska, I, Kamp, M, Appice, A, Loglisci, C, Antonie, L, Zimmermann, A, Guidotti, R, and Ozgobek, O
- Subjects
Multi target ,Computer science ,business.industry ,Multi-task learning ,Regression analysis ,Artificial intelligence ,Regression algorithm ,Machine learning ,computer.software_genre ,business ,computer ,Field (computer science) - Abstract
Multi-target models are machine learning models that simultaneously predict several target attributes. Due to a high number of real-world applications, the field of multi-target prediction is actively developing. With the growing number of multi-target techniques, there is a need for comparing them among each other. However, while established procedures exist for comparing conventional, single-target models, little research has been done on making such comparisons in the presence of multiple targets. In this paper, we highlight the challenges of evaluating multi-target models, focusing on multi-target regression algorithms. This paper reviews the common practice and discusses its shortcomings, indicating directions for future research. ispartof: pages:353-362 ispartof: Proceedings of the 2nd Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning vol:1323 pages:353-362 ispartof: 2nd Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning at European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Database (ECML PKDD 2020) location:Gent, Belgium date:14 Sep - 18 Sep 2020 status: accepted
- Published
- 2020