A Comparision Between Membership function And Entropy Function In Fuzzy Adaptive Linear Regression

Abstract

There are many uncertainty sources that may affect the statistical reasoning. However, traditional methods can not deal with all kinds of uncertainty sources, which has led many researchers to develop traditional methods. Studies still exist to this day, making hypotheses to create a common understanding for the purpose of reaching new solutions through the use of new methods that combine traditional and modern theories of sources of uncertainty The aim of current study was to develop the adaptive fuzzy linear regression model in the case of using inaccurate data as the source of uncertainty. Specifically, the model proposed by [1]. However, instead of what dominant in fuzzy linear regression analysis, we used a new born method that uses the positions and entropy to fuzzification instead membership function. As for the comparison method we used the mean absolute difference as performance's accuracy measures. The results of this study showed the efficiency of the use of the position and the entropy function to describe the fuzzy numbers over the use of the membership functions. The results also indicated that the develop model has the best results compared to the model adapted using the membershop functions in [1].