Weather Forecasting Using ANFIS and ARIMA MODELS

Abstract

Abstract

This paper presents a comparative study of statistical and neuro-fuzzy network models for forecasting the weather of Göztepe, İstanbul, Turkey. For developing the models, we used nine year’s data (2000-2008) comprising of daily average temperature (dry-wet), air pressure, and wind-speed. We used Adaptive Network Based Fuzzy Inference System ( ANFIS  ) and ARIMA models. To ensure the effectiveness of ARIMA and  ANFIS    techniques, we also tested the different models using a different training and test data set. The criteria of performance evaluation are calculated in order to evaluate and compare the performances of ARIMA and  ANFIS    models. Hence, paper briefly explains how neuro-fuzzy models could be formulated using different learning methods and then investigates whether they can provide the required level of performance, which are sufficiently good and robust to provide a reliable model for practical weather forecasting. From the results, the best fit model and network structure are determined according to prediction performance and the approach is effective and reliable. The performance comparisons of  ANFIS    and ARIMA models due to  MAE,RMSE,R²  criteria, the  ANFIS   gives better results have been observed.