Anang Kurnia Hidayat, Arie Wahyu Wijayanto

Forecasting Waste Generation Volume Using Google Trends Index with the SARIMAX Method

Introduction

Forecasting waste generation volume using google trends index with the sarimax method. Forecast waste generation in Yogyakarta using Google Trends Index & SARIMAX. Accurate model (MAPE 5.7873) predicts future trends, informing adaptive waste management policies.

Abstract

In recent years, the Special Region of Yogyakarta has faced a growing challenge of waste generation exceeding its management capacity. This situation underscores the urgency of developing a long-term, data-driven waste management strategy. This study aims to build an accurate forecasting model for waste volume using real-time data from the Google Trends Index (GTI) alongside official statistical data as exogenous variables. The forecasting methods employed are SARIMA and SARIMAX, tested with various parameter and variable combinations. The best-performing model is SARIMAX(1,1,1)(1,0,0)12 with the Production Index (IBS) and the GTI for the keyword “sampah” (waste) as exogenous variables, achieving a MAPE of 5.7873 (classified as very good) and an RMSE of 46.7509. The forecast shows an upward trend in mid-2024, a decline at the end of 2024, and a sharp increase in early 2025. These results can inform adaptive waste management policies, particularly in strengthening upstream strategies such as waste reduction, sorting, and recycling.

Review

This study tackles the pressing issue of escalating waste generation in the Special Region of Yogyakarta by proposing an innovative approach to forecasting waste volume. The integration of the Google Trends Index (GTI) as an exogenous variable alongside traditional statistical data is a particularly timely and valuable contribution, offering a novel way to leverage real-time, high-frequency public interest data for environmental management. The paper's objective to build an accurate, data-driven forecasting model is highly relevant, especially given the urgent need for long-term, adaptive waste management strategies in the region. Methodologically, the choice of SARIMA and SARIMAX models is well-suited for time-series data with potential seasonality, commonly observed in waste generation patterns. The identified best-performing model, SARIMAX(1,1,1)(1,0,0)12, effectively combines the Production Index (IBS) and the GTI for the keyword "sampah" (waste), demonstrating a robust predictive capability with a reported MAPE of 5.7873 and an RMSE of 46.7509. These metrics affirm the model's "very good" accuracy, providing a solid foundation for practical application. The forecasted trends, including an upward surge in mid-2024 and a sharp increase in early 2025, offer critical foresight that can directly inform the development and strengthening of upstream waste management policies, such as reduction, sorting, and recycling initiatives. While the abstract clearly highlights the study's strengths and promising results, a complete manuscript would benefit from a more detailed discussion on the selection process of the GTI keywords and a sensitivity analysis of the model to alternative keywords or combinations. Furthermore, exploring the potential limitations and nuances of using public search interest as a proxy for physical waste generation, and how these factors might influence long-term forecast reliability, would enhance the study's robustness. Nonetheless, this research presents a significant step forward in applying advanced econometric models and novel data sources to address a crucial urban challenge, offering a valuable blueprint for other regions facing similar waste management pressures.

Full Text

Comments

You need to be logged in to post a comment.