Contents lists available at ScienceDirect Transportation Research Part C journal homepage: www.elsevier.com/locate/trc A context aware system for driving style evaluation by an ensemble learning on smartphone sensors data Mohammad Mahdi Bejani, Mehdi Ghatee ⁎ Department of Computer Science, Amirkabir University of Technology, Hafez Ave., Tehran 15875-4413, Iran ARTICLE INFO Keywords: Driving style evaluation Sensor fusion Feature extraction Ensemble machine learning algorithms Fuzzy inference system ABSTRACT There are many systems to evaluate driving style based on smartphone sensors without enough awareness from the context. To cover this gap, we propose a new system namely CADSE system to consider the effects of traffic levels and car types on driving evaluation. CADSE system in- cludes three subsystems to calibrate smartphone, to classify the maneuvers, and to evaluate driving styles. For each maneuver, the smartphone sensors data are gathered in three successive time intervals referred as pre-maneuver, in-maneuver, and post-maneuver times. Then, we ex- tract some important mathematical and experimental features from these data. Afterwards, we propose an ensemble learning method on these features to classify the maneuvers. This ensemble method includes decision tree, support vector machine, multi-layer perceptron, and k-nearest neighbors. Finally, we develop a rule-based fuzzy inference system to integrate the outputs of these algorithms and to recognize dangerous and safe maneuvers. CADSE saves this result in driver’s profile to consider more for dangerous driving recognition. The experimental results show that accuracy, precision, recall, and F-measure of CADSE system are greater than 94%, 92%, 92%, and 93%, respectively that prove the system efficiency. 1. Introduction The dangerous driver refers to a driver who drives with some dangerous maneuvers, which could be caused an accident. Such driver may be recognized by insurance companies or police offices to limit the dangerous driving and to improve the drivers’ behavior. Since, evaluation of drivers’ maneuvers is an important task in the subject of transportation safety, many issues were considered for this problem, see e.g., Tijerina (2000). For this problem, Toledo et al. (2008) focused on data gathering processes. Duddle and Perham (2007), Raz et al. (2008), McCall and Trivedi (2006), and Olaverri-Monreal et al. (2010) developed different methods and systems for evaluating the quality of maneuvers and driving style recognition. The effect of such evaluation systems for driving profile was also mentioned by Raz et al. (2009). Also for insurance companies McMillan et al. (2000) and Raz et al. (2006) developed some driving evaluation applications. On the other hand, the quality of maneuvers depends on changes of acceleration and the order of operations. Xu et al. (2015) showed that the lateral accelerations of vehicles can be converted into the lateral force coefficient, which is a key factor for vehicle lateral stability, driving safety and driving behavior. Usually a dangerous maneuver relates to high acceleration, but not at all. In the other words, some dangerous maneuvers happen under low acceleration; For example, when a driver changes his lane exactly before his U-Tern, the maneuver is dangerous even if the acceleration is low. This means the order of the operations is important for https://doi.org/10.1016/j.trc.2018.02.009 Received 7 September 2017; Received in revised form 7 February 2018; Accepted 14 February 2018 ⁎ Corresponding author. E-mail address: ghatee@aut.ac.ir (M. Ghatee). URL: http://www.aut.ac.ir/ghatee (M. Ghatee). Transportation Research Part C 89 (2018) 303–320 0968-090X/ © 2018 Elsevier Ltd. All rights reserved. T