Predicting and Identifying the Finest Model for Forecasting Maximum Temperature in Rajshahi District using Machine Learning Approaches

Abstract

Temperature plays a significant role in driving climate change. Climate change tries to distinguish changes within the cruel values of temperature arrangement over time. Compared to other locales the most extreme temperature of Rajshahi City is high in the summer season and low in the winter season. The main objective of this paper was to identify the best model for forecasting maximum temperature in Rajshahi. Firstly, we determined classical ARIMA model then applied seven machine learning based algorithms like as support vector machine(SVM), Bayesian Regularized Neural networks(BRNN), Neural Network(Nnet), Elastic Net(ENet), Least Absolute Shrinkage and Selection Operation(Lasso), Relevance Vector Machines With Linear Kernel(RVML) for forecasting temperature data. Machine learning (ML) based models were upgraded utilizing arbitrary look with tune length 1000 and bootstrap-based 10 time’s cross-validation. In this paper, we considered the month-to-month greatest temperature forecast in Bangladesh's Rajshahi division. We utilized a set of classical and machine learning based approaches strategies to anticipate the most extreme temperature of the Rajshahi locale. Although other machine learning like ARIMA performed great too this result demonstrated that the ENet showed more proficient and way better than other models. The values of MAE, RMSE, MAPE and R² are 0.97, 1.29, 2.74 and 89%. Hence the comes about such as execution of the forecasting model may be progressed by utilizing ENet compared to routine ARIMA & other models, particularly for most extreme temperature expectations. This investigation can be useful for organizations mindful of checking climate and defining arrangements.

IJSS, Vol. 26(1), March, 2026, pp 119-134