Soft computing models to predict daily temperature of Dhaka

Abstract

Soft computing forecasting tools play an important role to forecast many complicated systems. In this paper, an effort has been made to use soft computing approaches to predict Dhaka daily temperatures for the period of 28 February 1945 to 27 August 2006. We have selected the fuzzy neuro model, the neuro genetic algorithm model as soft computing techniques. To compare results, a popular time series statistical technique, namely autoregressive moving average model is selected and based on error analysis, a suitable model to predict temperature for Dhaka city is proposed. The performance comparisons of different models due to root mean square error, correlation coefficient and coefficient of determination between observed and predicted temperatures indicate that the neuro genetic algorithm model predicts temperatures with maximum accuracy, followed by the fuzzy neuro model. Our believe findings of this paper will be useful for those who are interested about Bangladeshi important atmospheric parameter, namely temperature.