Application of soft computing techniques in modeling time series data

Abstract

Time Series is a sequence of equally spaced observation of a variable in a particular domain. The newlineobjective of this research is to develop time series models using soft computing techniques which newlineare capable of forecasting future events efficiently. Fuzzy time series models deal with uncertain newlineand imprecise information efficiently. Research on developing Fuzzy Time Series Forecasting newlinemodels is gaining momentum these days. Certain features of this model make it more suitable newlinethan the classical models. Observations of fuzzy time series are fuzzy sets and a relationship newlineexists between the observations at present time and those at previous times. It is to be noted that newlinea fuzzy time series is not fuzzy. newlineThere are several on-going efforts to improve the performance of the model. Ultimate goal is to newlineminimize forecast accuracy. Working in that direction, we intend to find out flaw in fuzzy time newlineseries model designing and provide possible solution to improve the forecast accuracy by newlinedeveloping new models for forecasting fuzzy time series. Studying basic structure of the models newlineit can be said that the performance of the models is dependent on two major phases: (1) Data newlinepartitioning/Interval creation technique and (2) Establishing rules for Prediction/ Forecasting. newline