Driving Behavior Analysis Using Deep Learning on GPS Data

Aggressive drivers are often considered to violate traffic rules and adopt dangerous driving behavior. This requires the development of effective and robust classifiers for unsafe drivers. Driving behavior analysis is the classification of driving behavior based on the driver’s GPS trajectory. With ever-increasing GPS trajectory data, dangerous driving behavior can be thoroughly analyzed and better classified using a deep learning model. Behavioral analytics can help us analyze and identify dangerous drivers that contribute to traffic safety and promote safe driving behavior. In this paper, we propose a novel feature extraction model using a statistical approach to extract the important features from the GPS trajectory data and label the trajectory. To overcome the dataset dependency, we propose to use a deep learning model on our labeled data and finally classify the safe and unsafe drivers. The proposed method demonstrates high accuracy with reduced computational overhead.