The Proposed Data-Driven glassoFCM Method

Abstract

Fuzzy cognitive maps (FCMs) have gained popularity within the scientific community due to their capabilities in modeling and decision making for complex problems. However, learning FCM models automatically from data without any expert knowledge and/or historical data remains a considerable challenge. Therefore, we propose a novel algorithm, the glassoFCM (graphical least absolute shrinkage and selection operator and Fuzzy Cognitive Map) to bridge that gap in the literature. Specifically, the glassoFCM is a combination of glasso technique with EBIC (Extended Bayesian Information Criterion) regularization and the FCM methodology. To recap, the glasso is a technique originated from machine learning to model data structures or in simple word to learn the undirected structure of a Gaussian Graphical Model and the FCM method is used to simulate different decision-making scenarios.