The Links between Statistical and Fuzzy Models for Time Series Analysis and Forecasting

Abstract

Traditionally, time series have been a study object for Statistics. A number of models and techniques have been developed within the field to cope with time series of increasing difficulty. On the other hand, fuzzy systems have been proved quite effective in a vast area of applications. Researchers and practitioners quickly realized that time series could also be approached with fuzzy and other soft computing techniques. Unfortunately, for a long time both communities have somehow ignored each other, disregarding interesting results and procedures developed in the other area.We addressed the problem of digging in the links between Statistical and fuzzy models for time series analysis and forecasting. In this chapter we present some of the most relevant results we have found in this area. In particular we introduce a new procedure based on statistical inference to build fuzzy systems devoted to time series modelling.