Rado Yendra, Muhammad Marizal, Hilvania Ramadhani

Forecasting of Average Air Temperature in the City of Pekanbaru Using the Holt-Winters Method

Introduction

Forecasting of average air temperature in the city of pekanbaru using the holt-winters method. Forecast Pekanbaru's average air temperature (2017-2024 BMKG data) using the Holt-Winters method. Predict 2025 trends for climate change mitigation planning. MAPE 2.684.

Abstract

Global climate change causes significant fluctuations in air temperature, including in the city of Pekanbaru, therefore, a predictive system is needed that can help the government and the community in dealing with climate impacts, one of which is through air temperature forecasting. This study aims to forecast the average air temperature in Pekanbaru City using the Holt-Winters Exponential Smoothing method, which is known to be effective in capturing seasonal patterns and trends. The data used is monthly average air temperature data from 2017 to 2024 obtained from BMKG. The analysis was carried out using an addictive approach and model evaluation was carried out based on the Mean Absolute Percentage Error (MAPE) value. The results show that the best model is obtained on a parameter with a MAPE value of 2.684. This model is then used to forecast the air temperature in 2025, which is predicted to decrease gradually. The results of this forecast are expected to be a reference in planning and decision-making related to climate change mitigation in the Pekanbaru area

Review

This study addresses a highly relevant and pressing issue: forecasting average air temperature in Pekanbaru, Indonesia, in the context of global climate change. The authors appropriately selected the Holt-Winters Exponential Smoothing method, a well-regarded technique for time series data exhibiting trends and seasonality, to analyze monthly average air temperature data from 2017 to 2024 obtained from BMKG. The stated aim to provide a predictive system for local government and community planning aligns well with critical climate change adaptation efforts, highlighting the practical significance of this research. The methodology, employing an additive Holt-Winters approach and evaluating model performance using Mean Absolute Percentage Error (MAPE), is sound. The reported MAPE value of 2.684 for the best model suggests a very high degree of accuracy in capturing the historical patterns of Pekanbaru's air temperature, which is a commendable achievement. This accurate model is then used to forecast air temperature for 2025, predicting a "gradual decrease." While the accuracy is impressive, it would be beneficial for the full paper to provide context on whether other forecasting models were considered for comparison, to further solidify the choice and superiority of Holt-Winters for this specific dataset. Additionally, a deeper interpretation of the predicted "gradual decrease" in 2025, including its potential magnitude and the underlying factors the model might be capturing, would enhance the findings' impact. In conclusion, this research presents a valuable contribution by successfully applying a robust forecasting method to a critical environmental challenge in Pekanbaru. The high accuracy of the developed model offers a practical tool for informed decision-making and planning related to climate change mitigation and adaptation strategies in the region. The findings serve as a strong foundation for future work, which could explore incorporating additional climatic variables, extending forecasting horizons with appropriate uncertainty quantification, or developing stakeholder-friendly platforms to disseminate these vital predictions more broadly.

Full Text

Comments

You need to be logged in to post a comment.