Selection of Objective Measures for Ranking in Associative Classifiers

• Grant number: 22/11990-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2022
• End date: November 30, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Veronica Oliveira de Carvalho
• Grantee: Fernando Augusto Serafim
• Host Institution: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil

Abstract

Associative classifiers (ACs) are predictive models based on association rules (ARs). The algorithms of this family generate the models in three or four steps, the second of which is responsible for ranking the classification association rules (CARs), a special type of AR that contains a class as a consequent. This step is commonly performed using two objective measures (OMs), support and confidence. However, studies show that these measures may not be the most appropriate in different contexts. Besides, there are more than 60 OMs documented, which makes it difficult to choose. This proposal is linked to a master's project that aims, among other objectives, to "propose and/or adapt AC algorithms, aiming better models in terms of effectiveness, based on the aggregation of OMs". Currently, a pre-defined ("fixed") set of OMs is being used to perform the aggregation of the measures. Considering the above, this work aims to: (i) perform an analysis in relation to the types of measures that will be aggregated (symmetric; asymmetric; symmetric+asymmetric); (ii) explore ways to make the process of choosing OMs to be aggregated "dynamic" via feature selection methods.