Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

Importance Analysis in the Evaluation of Input Attributes of Classifiers

Elena Zaitseva1,a, Vitaly Levashenko1,b and Sergey Stankevich2

1Department of Informatics, University of Zilina, Slovakia.

2Scientific Centre for Aerospace Research of the Earth, NAS of Ukraine, Ukraine.


Most often, the techniques of Machine learning are used for the decision of problems in Reliability Analysis. In this study, we propose to consider the application of the Reliability Analysis based method application for the decision problem in Machine Learning, in particular, the analysis of the influence of input attributes on the classification result. Some attributes are most important for the classification because they significantly influence the classification result than others. A new method for the determination of the most important attributes is proposed. This method is developed based on the approach of Importance Analysis, which is widely used in Reliability Analysis. The attribute's importance is evaluated by structural importance.

Keywords: Classification, Importance analysis, Multi-state system, Structure function, attributes selection.

Download PDF