{"id":{"repo_id":"umn","oai_identifier":"oai:conservancy.umn.edu:11299/276715"},"canonical_url":"https://search.dev.ndltd.org/etd/umn/oai:conservancy.umn.edu:11299/276715","repository":{"repo_id":"umn","name":"University of Minnesota","base_url":"https://conservancy.umn.edu/server/oai/request"},"display":{"title":"A machine learning approach to predict emotional arousal and aggressive driving from EDA and heart rate signals","abstract":"Aggressive driving behaviors, such as sudden acceleration and hard braking, abrupt steering, significantly increase the risk of road accidents, especially among young drivers. This study aims to predict aggressive driving behavior and emotional arousal using physiological signals, Electrodermal Activity (EDA) and Heart Rate (HR), collected via the Empatica E4 wristband in real-world driving conditions. A total of 25 drivers participated in this study. From the collected Electrodermal activity (EDA) and Heart Rate signals, 22 features were extracted, including tonic and phasic components, difference-based features, and heart rate statistics. Feature selection was performed using Random Forest feature importance and L1-regularized Logistic Regression feature importance method to identify the most relevant features for three prediction tasks: emotional arousal, aggressive acceleration, and aggressive braking. These features were used to train four classification models, Random Forest, Logistic Regression, Support Vector Machine (SVM), and XGBoost, as well as an ensemble VotingClassifier. Model performance was evaluated using five-fold cross-validation across accuracy, precision, recall, and F1-score. The ensemble model achieved the best overall performance across tasks. For arousal prediction using logistic regression feature importance, the ensemble achieved an accuracy of 0.65, a precision of 0.69, a recall of 0.60, and an F1-score of 0.57. In aggressive acceleration prediction, the highest-performing model (XGBoost) achieved 0.62 accuracy, while ensemble models reached 0.55. For aggressive braking prediction, both SVM and the ensemble achieved 0.70 accuracy. These findings indicate that physiological signals can reliably predict emotional arousal and driving behaviour. This research highlights the potential of integrating wearable sensor data with machine learning to build real-time driver monitoring systems aimed at improving road safety.","abstract_html":"Aggressive driving behaviors, such as sudden acceleration and hard braking, abrupt steering, significantly increase the risk of road accidents, especially among young drivers. This study aims to predict aggressive driving behavior and emotional arousal using physiological signals, Electrodermal Activity (EDA) and Heart Rate (HR), collected via the Empatica E4 wristband in real-world driving conditions. A total of 25 drivers participated in this study. From the collected Electrodermal activity (EDA) and Heart Rate signals, 22 features were extracted, including tonic and phasic components, difference-based features, and heart rate statistics. Feature selection was performed using Random Forest feature importance and L1-regularized Logistic Regression feature importance method to identify the most relevant features for three prediction tasks: emotional arousal, aggressive acceleration, and aggressive braking. These features were used to train four classification models, Random Forest, Logistic Regression, Support Vector Machine (SVM), and XGBoost, as well as an ensemble VotingClassifier. Model performance was evaluated using five-fold cross-validation across accuracy, precision, recall, and F1-score. The ensemble model achieved the best overall performance across tasks. For arousal prediction using logistic regression feature importance, the ensemble achieved an accuracy of 0.65, a precision of 0.69, a recall of 0.60, and an F1-score of 0.57. In aggressive acceleration prediction, the highest-performing model (XGBoost) achieved 0.62 accuracy, while ensemble models reached 0.55. For aggressive braking prediction, both SVM and the ensemble achieved 0.70 accuracy. These findings indicate that physiological signals can reliably predict emotional arousal and driving behaviour. This research highlights the potential of integrating wearable sensor data with machine learning to build real-time driver monitoring systems aimed at improving road safety.","abstract_has_math":false,"creators":["Nahid, Md Sakibul Hasan"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-06","date_published":"2025-06","updated_at":"2026-07-24T05:19:52Z","subjects":["Aggressive Driving","Electrodermal Activity","Heart Rate","Machine Learning","Young Drivers","Plan As (thesis-based master's degrees)","Master of Science","Master of Science in Mechanical Engineering","Department of Mechanical and Industrial Engineering","Swenson College of Science and Engineering","University of Minnesota Duluth"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11299/276715","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Nahid, Md Sakibul Hasan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-10-24T11:32:59Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-10-24T11:32:59Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-06"]},{"key":"dc:type","label":"Dc Type","values":["Thesis or Dissertation"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Aggressive Driving","Electrodermal Activity","Heart Rate","Machine Learning","Young Drivers","Plan As (thesis-based master's degrees)","Master of Science","Master of Science in Mechanical Engineering","Department of Mechanical and Industrial Engineering","Swenson College of Science and Engineering","University of Minnesota Duluth"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11299/276715"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["University of Minnesota M.S.M.E. thesis. June 2025. Major: Mechanical Engineering. Advisor: Turuna Seecharan. 1 computer file (PDF); xi, 117 pages."]},{"key":"dc:description.abstract","label":"Abstract","values":["Aggressive driving behaviors, such as sudden acceleration and hard braking, abrupt steering, significantly increase the risk of road accidents, especially among young drivers. This study aims to predict aggressive driving behavior and emotional arousal using physiological signals, Electrodermal Activity (EDA) and Heart Rate (HR), collected via the Empatica E4 wristband in real-world driving conditions. A total of 25 drivers participated in this study. From the collected Electrodermal activity (EDA) and Heart Rate signals, 22 features were extracted, including tonic and phasic components, difference-based features, and heart rate statistics. Feature selection was performed using Random Forest feature importance and L1-regularized Logistic Regression feature importance method to identify the most relevant features for three prediction tasks: emotional arousal, aggressive acceleration, and aggressive braking. These features were used to train four classification models, Random Forest, Logistic Regression, Support Vector Machine (SVM), and XGBoost, as well as an ensemble VotingClassifier. Model performance was evaluated using five-fold cross-validation across accuracy, precision, recall, and F1-score. The ensemble model achieved the best overall performance across tasks. For arousal prediction using logistic regression feature importance, the ensemble achieved an accuracy of 0.65, a precision of 0.69, a recall of 0.60, and an F1-score of 0.57. In aggressive acceleration prediction, the highest-performing model (XGBoost) achieved 0.62 accuracy, while ensemble models reached 0.55. For aggressive braking prediction, both SVM and the ensemble achieved 0.70 accuracy. These findings indicate that physiological signals can reliably predict emotional arousal and driving behaviour. This research highlights the potential of integrating wearable sensor data with machine learning to build real-time driver monitoring systems aimed at improving road safety."]},{"key":"dc:title","label":"Title","values":["A machine learning approach to predict emotional arousal and aggressive driving from EDA and heart rate signals"]}]}],"canonical_facts":{"dc:creator":["Nahid, Md Sakibul Hasan"],"dc:date.accessioned":["2025-10-24T11:32:59Z"],"dc:date.available":["2025-10-24T11:32:59Z"],"dc:date.issued":["2025-06"],"dc:description":["University of Minnesota M.S.M.E. thesis. June 2025. Major: Mechanical Engineering. Advisor: Turuna Seecharan. 1 computer file (PDF); xi, 117 pages."],"dc:description.abstract":["Aggressive driving behaviors, such as sudden acceleration and hard braking, abrupt steering, significantly increase the risk of road accidents, especially among young drivers. This study aims to predict aggressive driving behavior and emotional arousal using physiological signals, Electrodermal Activity (EDA) and Heart Rate (HR), collected via the Empatica E4 wristband in real-world driving conditions. A total of 25 drivers participated in this study. From the collected Electrodermal activity (EDA) and Heart Rate signals, 22 features were extracted, including tonic and phasic components, difference-based features, and heart rate statistics. Feature selection was performed using Random Forest feature importance and L1-regularized Logistic Regression feature importance method to identify the most relevant features for three prediction tasks: emotional arousal, aggressive acceleration, and aggressive braking. These features were used to train four classification models, Random Forest, Logistic Regression, Support Vector Machine (SVM), and XGBoost, as well as an ensemble VotingClassifier. Model performance was evaluated using five-fold cross-validation across accuracy, precision, recall, and F1-score. The ensemble model achieved the best overall performance across tasks. For arousal prediction using logistic regression feature importance, the ensemble achieved an accuracy of 0.65, a precision of 0.69, a recall of 0.60, and an F1-score of 0.57. In aggressive acceleration prediction, the highest-performing model (XGBoost) achieved 0.62 accuracy, while ensemble models reached 0.55. For aggressive braking prediction, both SVM and the ensemble achieved 0.70 accuracy. These findings indicate that physiological signals can reliably predict emotional arousal and driving behaviour. This research highlights the potential of integrating wearable sensor data with machine learning to build real-time driver monitoring systems aimed at improving road safety."],"dc:identifier.uri":["https://hdl.handle.net/11299/276715"],"dc:language.iso":["en"],"dc:subject":["Aggressive Driving","Electrodermal Activity","Heart Rate","Machine Learning","Young Drivers","Plan As (thesis-based master's degrees)","Master of Science","Master of Science in Mechanical Engineering","Department of Mechanical and Industrial Engineering","Swenson College of Science and Engineering","University of Minnesota Duluth"],"dc:title":["A machine learning approach to predict emotional arousal and aggressive driving from EDA and heart rate signals"],"dc:type":["Thesis or Dissertation"]},"updated_at":"2026-07-24T05:19:52Z"}