An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization

In order to provide safe development of road freight traffic, this paper proposes a truck driving risk identification method based on Optuna optimization of machine learning model. First, the risk characterization indicators were extracted from the natural driving data of trucks, and the threshold v...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhaofei Wang, Hao Li, Qiuping Wang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10909099/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850070640402366464
author Zhaofei Wang
Hao Li
Qiuping Wang
author_facet Zhaofei Wang
Hao Li
Qiuping Wang
author_sort Zhaofei Wang
collection DOAJ
description In order to provide safe development of road freight traffic, this paper proposes a truck driving risk identification method based on Optuna optimization of machine learning model. First, the risk characterization indicators were extracted from the natural driving data of trucks, and the threshold value of each indicator was determined using a box plot-based method. Second, the truck driving risk was quantified into three categories of low level, medium level, and high level risk, and the unbalanced data were processed using a hybrid sampling algorithm. Finally, the tree-based decision tree (DT) model, random forest (RF) model, Light Gradient Boosting Machine (LightGBM) model and eXtreme Gradient Boosting (XGBoost) model were selected for training and Optuna was used for hyperparameter optimization of the model. The results are shown to indicate that the machine learning model based on Optuna optimization can effectively identify truck driving risks. Combining the running time, precision, recall, and F1-Score, the LightGBM model optimized based on the Tree-structured Parzen Estimator (TPE) algorithm has the best performance with a precision of 0.98. In addition, the speed mean has the highest feature importance of 14%, which needs to be focused on when preventing truck driving risks. The research results can provide policy support for transportation management departments to formulate risk control measures for trucks.
format Article
id doaj-art-fe42e69d37a24d3d9195126b99abd1f5
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-fe42e69d37a24d3d9195126b99abd1f52025-08-20T02:47:29ZengIEEEIEEE Access2169-35362025-01-0113427234273210.1109/ACCESS.2025.354744510909099An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna OptimizationZhaofei Wang0https://orcid.org/0000-0002-5070-4917Hao Li1https://orcid.org/0009-0000-9240-3234Qiuping Wang2College of Urban Development and Modern Transportation, Xi’an University of Architecture and Technology, Xi’an, ChinaCollege of Urban Development and Modern Transportation, Xi’an University of Architecture and Technology, Xi’an, ChinaCollege of Urban Development and Modern Transportation, Xi’an University of Architecture and Technology, Xi’an, ChinaIn order to provide safe development of road freight traffic, this paper proposes a truck driving risk identification method based on Optuna optimization of machine learning model. First, the risk characterization indicators were extracted from the natural driving data of trucks, and the threshold value of each indicator was determined using a box plot-based method. Second, the truck driving risk was quantified into three categories of low level, medium level, and high level risk, and the unbalanced data were processed using a hybrid sampling algorithm. Finally, the tree-based decision tree (DT) model, random forest (RF) model, Light Gradient Boosting Machine (LightGBM) model and eXtreme Gradient Boosting (XGBoost) model were selected for training and Optuna was used for hyperparameter optimization of the model. The results are shown to indicate that the machine learning model based on Optuna optimization can effectively identify truck driving risks. Combining the running time, precision, recall, and F1-Score, the LightGBM model optimized based on the Tree-structured Parzen Estimator (TPE) algorithm has the best performance with a precision of 0.98. In addition, the speed mean has the highest feature importance of 14%, which needs to be focused on when preventing truck driving risks. The research results can provide policy support for transportation management departments to formulate risk control measures for trucks.https://ieeexplore.ieee.org/document/10909099/Traffic safetytruck driving riskmachine learningOptuna optimization
spellingShingle Zhaofei Wang
Hao Li
Qiuping Wang
An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
IEEE Access
Traffic safety
truck driving risk
machine learning
Optuna optimization
title An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
title_full An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
title_fullStr An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
title_full_unstemmed An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
title_short An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization
title_sort approach to truck driving risk identification a machine learning method based on optuna optimization
topic Traffic safety
truck driving risk
machine learning
Optuna optimization
url https://ieeexplore.ieee.org/document/10909099/
work_keys_str_mv AT zhaofeiwang anapproachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization
AT haoli anapproachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization
AT qiupingwang anapproachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization
AT zhaofeiwang approachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization
AT haoli approachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization
AT qiupingwang approachtotruckdrivingriskidentificationamachinelearningmethodbasedonoptunaoptimization