A Method for the Fuzzification of Categorical Variables
E. Jodoin, Carlos Andrés Peña-Reyes, Eduardo García Sánchez · 2006
Besides the numeric variables which are common in fuzzy modeling, some variables involved in the description of specific behaviors are categorical. Such variables are discrete, have no order a-priori, and most of the time handle a large amount of values (e.g., genes, proteins, countries, religions, etc.). This paper proposes a methodology for the fuzzification of categorical variables which could make part of a larger fuzzy modeling approach. The proposed solution allows to automatically create fuzzy membership functions for nominal categorical variables. We study some parameters so as to better assess their possible effect on the final outcome of the whole fuzzy modeling process.