Comparative Analysis of Machine Learning Models for CO Emission Prediction in Engine Performance - Research Article

Aug 5, 2026

This study presents a comparative analysis of machine learning models for predicting carbon monoxide (CO) emissions in automotive engines. Linear Regression, Decision Tree, Random Forest, and Support Vector Regression were trained and evaluated using engine performance parameters and measured emission data. Random Forest achieved the highest predictive accuracy with an R² of 0.8965. Feature importance results highlighted NOX, engine speed (RPM), and hydrocarbons (HC) as key predictors. Learning curve analysis supports generalization evaluation and indicates model limitations. The study supports cleaner, more efficient vehicle design.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Emission challenges and regulations

Machine learning opportunities

Research gaps

Study aims and evaluated models

Model set and objectives

|

SAKARYA UNIVERSITY JOURNAL OF COMPUTER AND
INFORMATION SCIENCES
[http://saucis.sakarya.edu.tr/](http://saucis.sakarya.edu.tr/)
e-ISSN: 2636-8129
Publisher: Sakarya University
Vol. 8, No. 1, 1-11, 2025 DOI: 10.35377/saucis...1564937 | | | --- | --- | | RESEARCH ARTICLE | | Comparative Analysis of Machine Learning Models for CO Emission Prediction in Engine Performance Beytullah Eren1, İdris Cesur2 1Sakarya University, Faculty of Engineering, Department of Environmental Engineering, Sakarya, Türkiye, [ror.org/04ttnw109](ror.org/04ttnw109) 2 Sakarya University of Applied Sciences, Faculty of Technology, Department of Mechanical Engineering, Sakarya, Türkiye, [ror.org/01shwhq58](ror.org/01shwhq58) | Corresponding author:
Beytullah Eren, Sakarya University,
Faculty of Engineering,
Department of Environmental Engineering [beren@sakarya.edu.tr](beren@sakarya.edu.tr) |
ABSTRACT
This study presents a comparative analysis of machine learning models for predicting carbon monoxide (CO) emissions in automotive engines. Four models—Linear Regression, Decision Tree, Random Forest, and Support Vector Regression—were evaluated using a dataset of engine performance parameters and emission measurements. Among these, the Random Forest model demonstrated the highest predictive accuracy, achieving an R² score of 0.8965. Feature importance analysis identified nitrogen oxides (NOX), engine speed (RPM), and | | --- | --- | | hydrocarbons (HC) as the most significant predictors of carbon monoxide emissions. Learning curve analysis | | | Article History:
Received: 10.10.2024
Revised: 17.12.2024
Accepted: 15.01.2025
Published Online: 27.03.2025 |
provided insights into model generalization and highlighted potential limitations. The study underscores the value of data-driven approaches in optimizing engine design and controlling emissions. The findings contribute to the development of cleaner, more efficient vehicles, supporting sustainability efforts in the automotive industry. This research bridges data science and automotive engineering, offering a framework for advanced emission prediction and control that can be applied to other pollutants and engine types.
Keywords: Carbon monoxide emissions, Machine learning, Random Forest, Engine performance optimization, Emission control, Sustainability, Automotive engineering | 1. Introduction The automotive industry is at a critical crossroads, tasked with enhancing engine performance while significantly reducing harmful emissions. Among these emissions, carbon monoxide (CO) poses a considerable threat to both human health and the environment [1] . As a byproduct of incomplete combustion, CO can lead to severe respiratory issues and, at high concentrations, may even be life-threatening [2] . Moreover, CO contributes to ground-level ozone formation, a major component of smog that exacerbates air quality concerns [3] . To combat these challenges, stringent global regulations, such as the European Union’s Euro 6 standards and the United States’ Tier 3 regulations, have been implemented, driving the need for innovative emission reduction strategies [4, 5] . Traditional approaches to emission control in internal combustion engines, such as optimizing engine design and using aftertreatment systems, often involve trade-offs with engine performance and fuel efficiency [6, 7] . These methods also struggle to address the complex, non-linear interactions between engine parameters and emission outputs, highlighting the limitations of conventional techniques. Recent advancements in data analytics and machine learning (ML) have introduced new opportunities to tackle these challenges. ML techniques are particularly adept at modeling complex, non-linear relationships between variables, enabling more accurate emission predictions and optimized control strategies [8, 9]. Studies have applied machine learning algorithms to predict pollutants like NOX

FAQs

Which machine learning models are compared for CO emission prediction?

The study evaluates Linear Regression, Decision Tree, Random Forest, and Support Vector Regression using engine performance parameters and emission measurements.

How does Random Forest perform compared with other models?

Random Forest shows the highest predictive accuracy, reaching an R² score of 0.8965.

What features are identified as most important for CO emissions?

Feature importance analysis indicates NOX, engine speed (RPM), and hydrocarbons (HC) are the most significant predictors of CO emissions.

Aug 5, 2026

Upload by River Wang