Hybrid ARMAX-ANN Model for Temperature Forecasting

Abstract

Many researchers have been interested in improving known forecasting methods using several methods, including hybridizing time series models with one, two, or more models in order to obtain greater accuracy in predicting the data of the series to be predicted. Therefore, our research is an extension of the researchers who preceded us in order to provide forecasting methods that have an impact in the field in which they are used. Therefore, our study focused on the ARMAX model, as it is a model that combines time series models and a regression model, and it becomes more effective when it is hybridized with one of the artificial intelligence models (ANN) in order to improve forecasting results, which are considered important in decision-making and in all areas of life, including the average temperature in Basra Governorate. A comparison was made between the ARMAX models, the ANN model, and the ARMAX-ANN hybrid model in order to predict temperatures for the years (2024-2028), as it became clear that the hybrid model is better at forecasting using the MSE criterion.