Browsing by Author "Tila, Tahrim Zaman"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    EDA Driving Data and Survey Responses
    (2023-04-05) Seecharan, Turuna; Tila, Tahrim Zaman; tseechar@d.umn.edu; Seecharan, Turuna; Gamification and Transportation Lab
    To find the relation between drivers’ stress levels and driving performance, two types of data were collected: drivers’ stress data, as known as Electrodermal Activity (EDA) Data, and vehicles’ engine data. The purpose of this study is to investigate how drivers’ driving performance changes in higher-stressed situations.
  • Thumbnail Image
    Item
    Exploring Relationship Between Electrodermal Activity and Driving Behavior
    (2023-08) Tila, Tahrim Zaman
    Driving is integral to many people's daily existence, but aggressive driving behavior increases the risk of road traffic collisions. Young, inexperienced drivers are more prone to aggressive driving and danger perception impairments. This thesis represents an investigation of the relationship between Electrodermal Activity (EDA) and engine data (acceleration, braking and speed). The research aims to explore how emotional arousal, as measured through EDA, impacts driving behavior, particularly in the context of acceleration, braking and speeding behaviors. The study utilizes 90 driving sessions, where EDA features, such as Tonic Mean, Tonic Max, and Phasic mean, are extracted from raw EDA data collected by E4 device, and engine data is collected to categorize acceleration and braking into Soft, Medium, and Hard classes. To capture complex and nonlinear relationships between EDA and engine data machine learning models were used. Three machine learning models, namely K-Nearest neighbors (KNN), Random Forest, and Support Vector Machine (SVM), are employed to predict emotional arousal, and driving behavior based on both raw EDA data and EDA features. The models are evaluated using performance metrics such as accuracy, precision, recall, and F1 score. EDA features works better with these models than raw EDA data in predicting arousal with an overall with accuracy of 58.14%, 66.67%, 74.44% and F1 score of 61%, 71.36%, 70.96% for KNN, Random Forest and SVM models respectively. In terms of predicting driving behavior, the SVM model performs almost similarly for both raw EDA and EDA features with overall accuracy of 67.63% and 73.59% respectively for acceleration and 72.03% and 63.17% respectively for braking. Overall, this thesis provides insights into the relationship between EDA and driving behavior. The findings have implications for designing interventions to promote safer driving practices based on individuals' emotional arousal and mood states.