Abstract
In this chapter, we elaborate on the construction of a FCM-based classifier for tabular data classification. The pipeline comprises exploratory data analysis, preliminary input processing, classification mechanism construction, and quality evaluation. The specifics of how to adapt an FCM to this task are discussed. We use a two-block FCM architecture. One block is specific to the input, and the second is used for class label generation. We have as many inputs as features and as many outputs as classes such that the weights are learned using Genetic Algorithms. The procedure is illustrated with a case study where we process a dataset named “wine”. The overall quality of a basic FCM-based classifier is shown, and the behavior of feature-related activation values is studied. The chapter contains a complete Python code for the elementary FCM-based classifier. The reader may conveniently follow and replicate the discussed experiment. Therefore, this chapter is specifically dedicated to those who wish to get well-acquainted with the elementary FCM-based classification model. The secondary goal of this chapter is to introduce notions essential to tabular data classification. These notions are utilized in the next chapters devoted to more advanced data classification models.
Original language | English |
---|---|
Title of host publication | Fuzzy Cognitive Maps |
Editors | Philippe J. Gianbbanelli, Gonzalo Nápoles |
Publisher | Springer Nature Switzerland AG |
Chapter | 9 |
Pages | 165-192 |
Number of pages | 28 |
ISBN (Print) | 978-3-031-48962-4 |
DOIs | |
Publication status | Published - 2024 |