{"title":"利用智能手机运动数据进行驾驶事件分类的灵敏度分析:分类器类型、传感器捆绑和数据采集率的情况","authors":"","doi":"10.1080/15472450.2022.2140048","DOIUrl":null,"url":null,"abstract":"<div><p>Classification of driving events is a crucial stage in driving behavior monitoring using smartphone sensory data. It has not been previously explored that to what extent classification performance depends on the classifier type and input data characteristics. To fill this gap, a real-world experiment is designed for supervised data collection. Then the effects of different machine learning (ML) classifiers, data sampling rates, and sensor combinations on the final classification accuracy are demonstrated. A considerable number of labeled events (4114) containing 11 types of driving maneuvers are collected using base sensors (accelerometer and gyroscope) and composite sensors (linear accelerometer and rotation vector) available in smartphones. Several models using 23 ML algorithms are trained. The sensitivity of these models is analyzed by changing the characteristics of the input data concerning the type of ML classifier, data sampling rate, and the bundle of mobile sensors. It is demonstrated that: (1) F1 scores vary from 70 to 96% for different ML classifiers, (2) F1 scores drop 30–40% depending on the classifier type when reducing the data sampling rate, and (3) using all four sensors as a bundle for classifying driving events is not reasonable since an approximate equal F1 score is achievable by a three-sensor bundle which includes an accelerometer and a linear accelerometer.</p></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"28 4","pages":"Pages 476-493"},"PeriodicalIF":2.8000,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Sensitivity analysis of driving event classification using smartphone motion data: case of classifier type, sensor bundling, and data acquisition rate\",\"authors\":\"\",\"doi\":\"10.1080/15472450.2022.2140048\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><p>Classification of driving events is a crucial stage in driving behavior monitoring using smartphone sensory data. It has not been previously explored that to what extent classification performance depends on the classifier type and input data characteristics. To fill this gap, a real-world experiment is designed for supervised data collection. Then the effects of different machine learning (ML) classifiers, data sampling rates, and sensor combinations on the final classification accuracy are demonstrated. A considerable number of labeled events (4114) containing 11 types of driving maneuvers are collected using base sensors (accelerometer and gyroscope) and composite sensors (linear accelerometer and rotation vector) available in smartphones. Several models using 23 ML algorithms are trained. The sensitivity of these models is analyzed by changing the characteristics of the input data concerning the type of ML classifier, data sampling rate, and the bundle of mobile sensors. It is demonstrated that: (1) F1 scores vary from 70 to 96% for different ML classifiers, (2) F1 scores drop 30–40% depending on the classifier type when reducing the data sampling rate, and (3) using all four sensors as a bundle for classifying driving events is not reasonable since an approximate equal F1 score is achievable by a three-sensor bundle which includes an accelerometer and a linear accelerometer.</p></div>\",\"PeriodicalId\":54792,\"journal\":{\"name\":\"Journal of Intelligent Transportation Systems\",\"volume\":\"28 4\",\"pages\":\"Pages 476-493\"},\"PeriodicalIF\":2.8000,\"publicationDate\":\"2024-07-03\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Intelligent Transportation Systems\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://www.sciencedirect.com/org/science/article/pii/S1547245023000294\",\"RegionNum\":3,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"TRANSPORTATION\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Intelligent Transportation Systems","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/org/science/article/pii/S1547245023000294","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"TRANSPORTATION","Score":null,"Total":0}
引用次数: 0
摘要
对驾驶事件进行分类是使用智能手机感知数据进行驾驶行为监控的关键阶段。分类性能在多大程度上取决于分类器类型和输入数据特征,这一点以前还没有人探讨过。为了填补这一空白,我们设计了一个用于监督数据收集的真实世界实验。然后展示了不同的机器学习(ML)分类器、数据采样率和传感器组合对最终分类准确性的影响。使用智能手机中的基本传感器(加速度计和陀螺仪)和复合传感器(线性加速度计和旋转矢量)收集了大量包含 11 种驾驶操作的标注事件(4114 个)。使用 23 种 ML 算法训练了多个模型。通过改变有关 ML 分类器类型、数据采样率和移动传感器捆绑的输入数据特征,分析了这些模型的灵敏度。结果表明(1) 对于不同的 ML 分类器,F1 分数从 70% 到 96% 不等;(2) 当降低数据采样率时,F1 分数会根据分类器类型下降 30%-40%;(3) 将所有四个传感器捆绑在一起对驾驶事件进行分类是不合理的,因为包括一个加速度计和一个线性加速度计在内的三个传感器捆绑在一起可以获得大致相同的 F1 分数。
Sensitivity analysis of driving event classification using smartphone motion data: case of classifier type, sensor bundling, and data acquisition rate
Classification of driving events is a crucial stage in driving behavior monitoring using smartphone sensory data. It has not been previously explored that to what extent classification performance depends on the classifier type and input data characteristics. To fill this gap, a real-world experiment is designed for supervised data collection. Then the effects of different machine learning (ML) classifiers, data sampling rates, and sensor combinations on the final classification accuracy are demonstrated. A considerable number of labeled events (4114) containing 11 types of driving maneuvers are collected using base sensors (accelerometer and gyroscope) and composite sensors (linear accelerometer and rotation vector) available in smartphones. Several models using 23 ML algorithms are trained. The sensitivity of these models is analyzed by changing the characteristics of the input data concerning the type of ML classifier, data sampling rate, and the bundle of mobile sensors. It is demonstrated that: (1) F1 scores vary from 70 to 96% for different ML classifiers, (2) F1 scores drop 30–40% depending on the classifier type when reducing the data sampling rate, and (3) using all four sensors as a bundle for classifying driving events is not reasonable since an approximate equal F1 score is achievable by a three-sensor bundle which includes an accelerometer and a linear accelerometer.
期刊介绍:
The Journal of Intelligent Transportation Systems is devoted to scholarly research on the development, planning, management, operation and evaluation of intelligent transportation systems. Intelligent transportation systems are innovative solutions that address contemporary transportation problems. They are characterized by information, dynamic feedback and automation that allow people and goods to move efficiently. They encompass the full scope of information technologies used in transportation, including control, computation and communication, as well as the algorithms, databases, models and human interfaces. The emergence of these technologies as a new pathway for transportation is relatively new.
The Journal of Intelligent Transportation Systems is especially interested in research that leads to improved planning and operation of the transportation system through the application of new technologies. The journal is particularly interested in research that adds to the scientific understanding of the impacts that intelligent transportation systems can have on accessibility, congestion, pollution, safety, security, noise, and energy and resource consumption.
The journal is inter-disciplinary, and accepts work from fields of engineering, economics, planning, policy, business and management, as well as any other disciplines that contribute to the scientific understanding of intelligent transportation systems. The journal is also multi-modal, and accepts work on intelligent transportation for all forms of ground, air and water transportation. Example topics include the role of information systems in transportation, traffic flow and control, vehicle control, routing and scheduling, traveler response to dynamic information, planning for ITS innovations, evaluations of ITS field operational tests, ITS deployment experiences, automated highway systems, vehicle control systems, diffusion of ITS, and tools/software for analysis of ITS.