An Expert System for Identification of Forecasting Model for Time Series

Abstract

<span style="font-family: 'Times New Roman','serif'; font-size: 12pt; mso-bidi-language: FA; mso-fareast-font-family: 'Times New Roman'; mso-ansi-language: EN-US; mso-fareast-language: EN-US;">Identifications </span><span style="font-family: 'Times New Roman','serif'; font-size: 11pt; mso-bidi-language: FA; mso-fareast-font-family: 'Times New Roman'; mso-ansi-language: EN-US; mso-fareast-language: EN-US;">and<br />analysis of time series are time consuming, based on trial and error and highly<br />dependent on expert judgments. This is mainly due to the presence of various<br />models for forecasting time series, as well as introducing new techniques for<br />analysis and predictions. In this paper, expert system structure is used to<br />replace traditional methods of model identifications for time series. Firstly,<br />several search engines are defined and analytical methods are specified. Next,<br />the knowledge base is developed such that a proper model can be assigned to<br />each data set. The goodness of fit is then evaluated by mathematical indices.<br />Repeating the process and modifying the responses to account for uncertain<br />situations, will provide a set of models to make the final decision. Lastly,<br />the performance of the proposed expert system is verified by a series of sample<br />data as a case study and the efficiency of the system is approved.</span>