Forecasting Knowledge Extraction by Computational Intelligence Techniques by Elia

Most of the recently developed intelligent systems have a knowledge base that incorporates the expertise domain knowledge and use it in reasoning chains during decision making. An important problem that should be solved by an intelligent system is forecasting the evolution of specific parameters that are monitored, for example. Among the various approaches that provide an efficient solution to the forecasting problems, some computational intelligence techniques allow the extraction of the forecasting knowledge under the IF-THEN rules form. The paper presents a general methodology that can be used to forecasting knowledge extraction and the experimental results of a comparative study between a computational intelligence technique, the adaptive neuro-fuzzy inference system (ANFIS), and a decision tree based technique, CART, applied to air pollution forecasting rules extraction, by following the proposed methodology.