The northeastern region of Bangladesh, particularly Sylhet, experiences considerable climatic problems characterized by higher rainfall, humidity, and temperature variations. Accurate forecasting of these meteorological variables is crucial for disaster resilience, public health, and socioeconomic stability. Conventional time‐series forecasting models, such as Seasonal Autoregressive Integrated Moving Average (SARIMA), could frequently fail to identify intricate, nonlinear trends. Artificial neural networks (ANNs) could also show lower forecasting accuracy when it comes to long‐term data. Hybrid models that integrate SARIMA with ANN provide improved forecasting accuracy and a more profound comprehension of climate trends. This study analyzed SARIMA, ANN, and hybrid SARIMA‐ANN models using univariate monthly average time‐series data of rainfall, temperature, and humidity for Sylhet. The dataset was obtained from the Sylhet Station of the Bangladesh Meteorological Department (BMD) and covered the period 1974–2022. Modeling accuracy was assessed using root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), coefficient of determination ( R 2 ), and Willmott’s index of agreement. The hybrid SARIMA‐ANN model consistently outperformed the SARIMA and ANN across all climatic variables. It achieved lower error measures and higher agreement indices. Projections for the period 2023–2032 indicated persistent climatic variability, demonstrating the model’s effectiveness in predicting climatic trends. The hybrid SARIMA‐ANN model provided a reliable framework for climate forecasting in Sylhet. Policymakers and agricultural planners are urged to implement this approach to reduce the socioeconomic effects of climatic unpredictability.
Shil et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: