Evaluation of Yager Families of Aggregation Operators in Discovering the Diatoms Indicating Properties

Abstract

Fuzzy aggregation operators perform operations between two fuzzy sets that satisfy certain axioms. They play an important role in fuzzy data mining process. As an integral part of many algorithms, the aggregation operators influence on the outcome of model and thus on the experimental evaluation of the models. Both pattern tree (PT) and the weighted pattern tree (WPT) algorithms use the aggregation operators to increase the accuracy of the model by making different operations between the descriptive and target attributes. Selecting the right operator is very important, especially considering generalized families of aggregation operators. Therefore, this paper aims to investigate the influence of the generalized Yager families of aggregation operators and their influence on both (PT and WPT) algorithms accuracy. This is done by modifying the λ parameter. This parameter is not the only parameter that influences the model performance, other factors are also in play, like the shape and the number of the membership functions (MFs), as well as the similarity metric. Our experimental evaluation will evaluate the descriptive and predictive performance of the models as well as the statistical significance of the results. The evaluation results show that the best descriptive and predictive models with both PT and WPT algorithms are obtained when λ is set to 1. For future work, we plan to investigate the influence of this family of aggregation operators with different similarity metrics, as well as other datasets.