THE PERFORMANCE ANALYSIS OF THE BEST MACHINE LEARNING MODEL FOR SULFUR DIOXIDE IN DKI JAKARTA

Aug 5, 2026

Upload by Noah

Clean air is essential for human health, and sulfur dioxide (SO2) is a key air pollutant influencing respiratory outcomes and environmental conditions such as acid rain. This study analyzes SO2 concentrations in DKI Jakarta over eleven years to identify the most accurate predictive model. It applies quantitative methods, machine learning techniques, and statistical evaluation. Three top-performing models are huber, exponential smoothing, and naive forecaster, with the naive forecaster achieving the best results by the reported error metrics.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Introduction

Background and problem focus

Literature Review

Air quality

Air components and pollutants

Sources of air pollution

Impacts on human health

Impacts on the environment

| p-ISSN 2338 – 3216 e-ISSN 2528 – 1070 | JURNAL STATISTIKA
UNIVERSITAS MUHAMMADIYAH SEMARANG
12(1) 2024: 34-47 | | | --- | --- | --- | THE PERFORMANCE ANALYSIS OF THE BEST MACHINE LEARNING MODEL FOR SULFUR DIOXIDE IN DKI JAKARTA Panji Kuswanaji1*, Rizky Addrian Aliyafi2, Agung Hari Saputra3 1 2 3 Undergraduate Applied Program of Meteorology, State College of Meteorology Climatology and Geophysics, Indonesia *[e-mail:](e-mail: pkuswanaji@gmail.com)[ pkuswanaji@gmail.com](e-mail: pkuswanaji@gmail.com) Article Info: Received: January 17, 2024 Accepted: May 30, 2024 Available Online: July 29, 2024 Keywords: Air Quality; Machine Learning; Model Abstract: A good clean air is one of crucial things for humans health. A place with good and clean air can prevent humans from various kinds of respiratory diseases. One of the factors that can influence the cleanliness of the air in an area is the composition of Sulfur Dioxide (SO2) . This study aims to examine sulfur dioxide (SO2) levels in Jakarta spanning eleven years, with the goal of determining the most accurate predictive model for SO2 concentrations., which is critical for public health and environmental management. The study incorporates quantitative methods, machine learning techniques, and statistical analysis. From this research there are three best models that has top performance, these are huber, exponential smoothing, and naive forecaster. The result shows that naive model has the best performance with MASE of 0.3864, RMSSE of 0.3098, MAE of 2.8857, RMSE of 3.7735, MAPE of 0.0593, and SMAPE of 0.0623. 1. INTRODUCTION A good clean air is one of crucial things for humans health. A place with good and clean air can prevent humans from various kinds of respiratory diseases. One of the factors that can influence the cleanliness of the air in an area is the level of contaminants or pollutants in an area [1] . Sulfur dioxide (SO2) is an air pollutant that causes coughing and shortness of breath. The main sources of SO2 in the air come from the combustion process (coal or diesel), the metallurgical industry, and the sulfuric acid industry [2] . SO2 gas is also the cause of acid rain and chemical photo fog which disrupts human life [3] . SO2 is a pollutant that generally comes from smoke or air waste that appears as a result of burning fuel [4] . Besides that, SO2 is also possible to come from smoke from large industries that have massive combustion processes and natural processes like volcanic eruption [4] . The smoke from the eruption contains SO2 and SO4 which can affect air quality and can even influence climate change [4] . In the realm of environmental monitoring and management, accurately forecasting air quality parameters is crucial for safeguarding public health and ensuring sustainable urban development. Sulfur dioxide (SO2), a significant pollutant with potential adverse effects on respiratory health and the environment, requires meticulous analysis, particularly in the dynamic urban landscape of DKI Jakarta [5] . Here, industrialization and urbanization are on KUSWANAJI, PANJI., ET AL the rise, making it imperative to comprehend and predict SO2 levels for effective pollution control and mitigation strategies. The focus in this study is to find the best performance machine learning model to predict SO2 levels in DKI Jakarta. In order to get that, this study compares various machine learning models such as naive forecaster [6], exponential smoothing [7], huber [8], extreme gradient boosting [9], and other PyCaret available models. The reason these models are used is because these models can forecast data based on previous data and can be trained to predict the pattern of future data. The best performance model in this study is the model which has the least error value (MASE, RMSSE, MAE, RMSE, MAPE, and SMAPE) . 2. LITERATURE REVIEW 2.1. Air Quality Air quality refers to atmospheric conditions that include the composition of ga

FAQs

Why is forecasting SO2 in DKI Jakarta important?

SO2 is a harmful air pollutant linked to respiratory problems and environmental effects. Forecasting it supports public health protection and pollution control strategies.

Which machine learning models performed best in the study?

The study reports three top models: huber, exponential smoothing, and naive forecaster. The naive forecaster shows the highest performance based on the listed error metrics.

How did the study evaluate model performance?

Model performance was assessed using error measures including MASE, RMSSE, MAE, RMSE, MAPE, and SMAPE. The model with the lowest error values is considered the best.

Aug 5, 2026

Upload by Noah