Bayesian network model for temperature forecasting in Dubai

Source of Publication

AIP Conference Proceedings


© 2018 Author(s). In this paper, we deal with the problem of weather forecasting using Bayesian networks. The study focuses on the data representing Dubai weather conditions. The variables used in this study are as follows: maximum and minimum temperature, mean temperature, mean relative humidity, rainfall, and wind speed. The National Centre of Meteorology and Seismology (NCMS) gathered the weather data from Dubai Airport Station located at latitude: 25°15' and longitude: 55°20' starting from January 2004 to December 2014. The values available represent month averages. We used these data to learn the Bayesian network structure and parameters. Inference in the Bayesian network helped in forecasting the maximum and minimum temperature of the succeeding months through dynamic Bayesian networks. The model showed 72% and 90% overall precision in forecasting minimum and maximum temperature, respectively.

Document Type

Conference Proceeding



Publication Date




Author First name, Last name, Institution

L. Smail, Zayed University