<p>The fuzzy regression functions approach is a fuzzy inference system based on fuzzy clustering method and multiple linear regression uses ordinary least squares. Since the fuzzy regression functions approach is based on the fuzzy clustering method, some of the inputs of this system are membership values. The intuitionistic fuzzy regression functions method, which is obtained by adapting the intuitionistic fuzzy clustering method, which is considered as a generalization of the fuzzy clustering method, to the fuzzy regression functions method, is a fuzzy inference system based on multiple linear regression just like the fuzzy regression functions method. The intuitionistic fuzzy regression functions method has more information than the fuzzy regression functions approach as it uses both membership, non-membership and hesitation degree as input. Both the fuzzy regression functions approach and the intuitionistic fuzzy regression functions approach have some problems due to the fact that they are based on multiple linear regression uses ordinary least squares. Ordinary least squares consider the linearity of the link between the independent and dependent variables and also can tend to overfitting if the model is too complicated. Unlike such features of ordinary least squares regression gaussian process regression is very useful in regression modelling where there are many different problems such as small sample size, high dimension and nonlinearity. From this point of view, an intuitionistic fuzzy regression functions based on Gaussian process regression is proposed in this study. The forecasting performance of this new proposed method is evaluated over various exchange time series. The proposed method is compared over various regression functions methods and it is concluded that the proposed method has superior forecasting performance.</p>