Machine learning models for predicting daily temperature extremes
Full Text (PDF)
Ahmet Ercan Topcu 1 , Mohammad Khaleel Ibrahim AlQallaf 1 , Yehia Ibrahim Alzoubi 2 , Ersin Elbasi 1 *
More Detail
1 College of Engineering and Technology, American University of the Middle East, KUWAIT2 College of Business Administration, American University of the Middle East, KUWAIT* Corresponding Author

Abstract

Accurate temperature prediction is a challenging task due to the complex and nonlinear nature of weather systems. Traditional statistical methods often struggle to capture these intricate relationships, leading to less reliable forecasts, especially in regions with diverse climatic conditions. The need for more advanced tools has driven the development of machine learning (ML) techniques. Hence, this study implemented and evaluated the performance of various models, including ridge regression, support vector regression (SVR), DT, RF, KNN, and neural network (NN). SVR attains the highest concordance correlation coefficient (CCC) 96% in South Korea, surpassing NN and RF 93%, while all models get an identical CCC 96% in Kuwait, demonstrating region-specific model effectiveness and data predictability for minimum temperatures. This suggests that NNs are well-suited for capturing complex patterns and relationships in temperature data. However, it is essential to note that model choice may vary depending on factors such as data quality, computational resources, and the desired level of interpretability. The process of model selection necessitates consideration of several practical trade-offs. Although the NN model attained the highest predictive accuracy, its training phase demanded significantly greater computational resources compared to SVR or RF. This study introduces a cross-regional comparison that reveals how climate and dataset complexity affect ML temperature prediction accuracy. Future research should quantitatively evaluate how specific climatic factors, like dryness, seasonal variations, and daily temperature, influence model efficacy, and investigate the integration of supplementary atmospheric and land-surface variables to enhance generalizability across various locations.

Keywords

License

This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Article Type: Research Article

EUR J SUSTAIN DEV RES, Volume 10, Issue 3, 2026, Article No: em0396

https://doi.org/10.29333/ejosdr/18318

Publication date: 01 Jul 2026

Online publication date: 06 Apr 2026

Article Views: 8

Article Downloads: 5

Open Access References How to cite this article