Predicting Konya's air temperature: Genetic programming, gradient boosting and random forest approaches

OpenAlex İngilizce

Average temperature prediction is important in many areas, such as climate change, agriculture, and energy management. It is also necessary for estimating energy demand, managing energy, and developing sustainable energy policies. In this study, using monthly average air temperature data between 1960-2017, temperature predictions were performed for Konya province using genetic programming, gradient boosting, and random forest techniques. The predicted average monthly temperature values between 2018-2021 were compared with the real values. Then, future predictions for the years 2022-2025 were also performed. Metrics such as R², RMSE, and MAE were used in model evaluations. R²=0.9477, RMSE=1.950 and MAE=1.500 for the genetic programming model, R²=0.9663, RMSE=1.564 and MAE=1.203 for the gradient boosting model, and R²=0.9905, RMSE=0.833 and MAE=0.625 for the random forest model. The same algorithms gave good results for future prediction of the average air temperature between 2022 and 2025. In conclusion, the applied machine learning methods gave successful results in monthly average air temperature predictions for Konya province, and these findings show that machine learning techniques can be used effectively in air temperature prediction.