Predicting Compliance with Speed Limits using Speed Limit Credibility Perception and Risk Perception Data

The link between attitudes and behavior shows that driving behavior can be predicted by personal characteristics and individual attitudes, as has been shown in previous research. This study aimed to predict the level of compliance with speed limits by individual drivers by using attitudes data including speed limit credibility perception and risk perception on eight rural single carriageway layouts. This study investigated how the road layout and roadside environment affect speed limit credibility perception and risk perception, and investigated which machine learning algorithm can be used to predict driving behavior based on experimental evidence. This study was carried out in a well-controlled experimental design by using a questionnaire and a driving simulator. The simulated road environment only considered rural single carriageway which has higher risk factors than other road types. The results show that a boosted decision tree algorithm can establish a driving behavior model based on drivers' credibility perception and risk perception. This result can be used to predict driving behavior in advance for in-vehicle warning system design.