Predicting Speeding Behavior of Long-Haul Freight Truck Drivers Using Machine Learning Models - Key Findings

Aug 4, 2026

Upload by Aurelia

Long-haul truck driver speeding behavior is shaped by weak enforcement of working-hour rules, tight deadlines, and heavy workloads that encourage sustained driving and risky decision-making. This study predicts speeding violations using statistical and machine learning models based on data from 370 respondents at two weigh stations in South Sulawesi, Indonesia, with eight socio-demographic, economic, and operational predictors. Binary Logistic Regression, Random Forest, and XGBoost are evaluated via accuracy, recall, F1-score, and AUROC, showing XGBoost as best performer.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Problem background and motivations

Prior research and gaps

Methods

Data collection and predictors

Model setup and evaluation metrics

Results

Model performance comparison

Key predictors of speeding violations

Discussion and implications

Safety insights for freight transportation

Predicting Speeding Behavior of Long-Haul Freight Truck Drivers Using Machine Learning Models Hakzah Hakzah 1*, Andi Damayanti 1 , Misbahuddin 1 , Abdul Rahman 1 1 Department of Civil Engineering, Universitas Muhammadiyah Parepare, Parepare, Indonesia. Received 14 August 2025; Revised 17 October 2025; Accepted 20 October 2025; Published 01 November 2025 Abstract The behavior of long-haul truck drivers is shaped by the weak enforcement of working-hour rules, tight deadlines, and heavy workloads. Over-dimensioning and overloading practices further increase risks by forcing drivers to handle excessive loads and work for prolonged periods. This study predicts speeding behavior among long-haul freight truck drivers using statistical and machine learning models. Data was collected from 370 respondents at two weigh stations in South Sulawesi, Indonesia, covering eight socio-demographic, economic, and operational predictors. Three models were tested: Binary Logistic Regression (BLR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) . The dataset was balanced and split into 70% training and 30% testing, with performance assessed using accuracy, recall, F1-score, and AUROC. XGBoost delivered the best results, achieving 97.3% accuracy, 93.2% recall, a 96.4% F1-score, and a perfect AUROC of 1.000. RF also showed strong performance with 94.05% accuracy and an AUROC of 0.973, while BLR served as a relevant baseline despite weaker predictions. Key predictors of speeding violations were daily sleep duration, monthly income, and driving experience. This study demonstrates how machine learning can be effectively integrated alongside transportation data under imbalanced conditions, providing evidence-based insights to strengthen freight transport safety. Keywords: Driver Behavior; Freight Transport; Speeding Violations; Machine Learning; XGBoost; Driver Safety. 1. Introduction Road freight transportation plays a crucial role in sustaining global logistics, particularly in developing economies where multimodal integration is still limited [1, 2]. Trucks continue to dominate long-distance distribution across varied terrains, yet the sharp growth of heavy vehicle traffic has intensified concerns regarding safety and operational risks. Prior studies have shown that human factors, including fatigue, stress, and economic pressure, remain major contributors to accidents involving freight trucks [3, 4] . Insufficient rest has frequently been linked to fatigue-related incidents in countries such as Australia [5], while payment-based incentives and strict delivery deadlines in North America have been associated with higher tendencies toward speeding and unsafe practices [6, 7] . These findings support theoretical perspectives on driver fatigue and risk-taking behavior, which emphasize that prolonged working hours and financial stress reduce attentiveness, impair reaction times, and increase the likelihood of risky decision-making. In Indonesia, systemic challenges exacerbate these risks. Weak regulation of driving hours, limited monitoring systems, and the persistent prevalence of over-dimension and overloading practices have been identified as conditions that amplify fatigue-induced behaviors [8, 9] . Drivers also frequently face socio-economic burdens, such as low wages and delivery pressures, which reinforce unsafe choices. According to behavioral economics theory, these structural stressors create conditions in which short-term financial gains are prioritized over long-term safety, thereby elevating the probability of traffic violations. * [Corresponding author: hakzah@umpar.ac.id](Corresponding author: hakzah@umpar.ac.id) [http://dx.doi.org/10.28991/CEJ-2025-011-11-015](http://dx.doi.org/10.28991/CEJ-2025-011-11-015) © 2025 by the authors. Licensee C.E.J, Tehran, Iran. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license ([http://creat

FAQs

What factors are described as influencing long-haul truck speeding behavior?

Weak working-hour enforcement, tight deadlines, and heavy workloads are highlighted, along with over-dimensioning and overloading practices that increase driver risk exposure and prolonged driving.

How was the dataset collected and what predictors were used?

Data were collected from 370 respondents at two weigh stations in South Sulawesi, covering eight socio-demographic, economic, and operational predictors.

Which machine learning model performed best for predicting speeding violations?

XGBoost achieved the strongest results, reaching 97.3% accuracy and an AUROC of 1.000, outperforming Random Forest and Logistic Regression.

Aug 4, 2026

Upload by Aurelia