Temperature Forecasting: A Comparison between Parametric and Non-Parametric Models

Authors

Alaa Sheta, Department of Computing Sciences, Texas A&M University-Corpus Christi, Corpus Christi, TX 78412, USAFollow
Ajay Katangur, Computer Science Department, Missouri State University, Springfield, MO 65897, USAFollow
Abdelkarim Baareh, Computer Science Department, Ajlune College, Al-Balqa Applied University, Ajlune, JordanFollow

Author Country (or Countries)

USA

Abstract

The development of accurate temperature prediction models is essential for not only human life but also for agricultural, animal life, tourism, and many others. Power consumption and achieving energy efficiency in buildings also depends on temperature. Although modeling-based regression is one of the most popular approaches, it still suffers from many difficulties related to the number of available measurements, the order of the model and the non-linearity of the data. In this paper, we provide a comparison between parametric and non-parametric models for temperature forecasting. We propose three-model structures to estimate the temperature in Mumbai, the business capital of India. They are parametric (i.e. Linear Regression (LR), Multi-gene Genetic Programming (MG-GP)) and non-parametric (i.e. Artificial Neural Networks (ANN)) models. These models are tested on data collected in Mumbai for the year of 2009. The results show that multi-gene GP model performs relatively well in predicting the temperature with a high degree of accuracy compared to the LR and ANN techniques.

Digital Object Identifier (DOI)

http://dx.doi.org/10.18576/amis/120604

Recommended Citation

Sheta, Alaa; Katangur, Ajay; and Baareh, Abdelkarim (2018) "Temperature Forecasting: A Comparison between Parametric and Non-Parametric Models," Applied Mathematics & Information Sciences: Vol. 12: Iss. 6, Article 4.
DOI: http://dx.doi.org/10.18576/amis/120604
Available at: https://digitalcommons.aaru.edu.jo/amis/vol12/iss6/4