Entropy-based fuzzy clustering and fuzzy modeling

摘要: Abstract Fuzzy clustering is capable of finding vague boundaries that crisp fails to obtain. But time complexity fuzzy usually high, and the need specify complicated parameters hinders its use. In this paper, an entropy-based method proposed. It automatically identifies number initial locations cluster centers. calculates entropy at each data point selects with minimum as first center. Next it removes all points having similarity larger than a threshold chosen This process repeated till are removed. Unlike previous methods kind, does not revise value for after center determined. saves lot time. Also requires just two easy specify. able find natural clusters in data. The also extended construct rule-based model. A new way estimating membership functions sets presented. experimental results show model good predicting output variable values.