Application of fuzzy regression models to predict exchange rates for composite currencies

This research proposes a new regression analysis model based on fuzzy statistics. With the increase in number of variables that interact in a complex economic environment, the accumulation of perfect knowledge for the purpose of prediction has become increasingly unrealistic. Thus, predicting future exchange rates in a composite currency situation has become increasingly difficult. Our research is an effort in that direction in which we try to predict certain key parameters based on the imperfect and uncertain information obtained from the related economic variables. In this regard, the theoretical foundation of fuzzy regression analysis has been extended. Here we utilize the fact that the relationship between the dependent variable and the independent variables is not sharply defined as in the non-fuzzy linear regression analysis. The most important assumption for this work is that the deviations between the estimated values and the corresponding real values of the output variables lie in the imprecision or the ambiguity in the system parameters. The significant contribution of this research is in its efficient modeling of fuzzy prediction analysis system which can be implemented in an uncertain economic environment such as chaotic fluctuations of composite currency.