Improving fuzzy systems identification with data transformations

A practical problem in the identification of fuzzy systems from data, is the design and the tuning of the membership functions. We demonstrate that if the data is properly transformed before the identification process, the resulting fuzzy model can be improved to the point it may not need a further tuning. The significance of the data transform can be validated using statistical methods. The method is demonstrated on a time series prediction problem, using the Box-Cox transform.