Investigation of Casual Factors on Traffic Accidents Using Mutual Information

In contemporary times, the prevalence of traffic accidents worldwide has escalated, prompting a pressing need to understand the causal factors contributing to these incidents. The concept of causal effect in the context of traffic accidents pertains to the influence or impact that specific variables or factors exert on the occurrence, severity, or outcome of accidents. Thus, it becomes paramount for researchers to delve into the root causes of these accidents. The primary challenge lies in identifying the causal relationships between various risk factors—such as weather conditions, human factors, and road conditions, among others—and the incidence of traffic accidents. Addressing this challenge forms the crux of this study. In this research endeavor, we propose a novel system that leverages mutual information-based quantification to elucidate the causal factors underlying traffic accidents. Furthermore, we employ machine learning techniques, such as support vector machines (SVM) and logistic regression, to predict the fatality of accidents. According to the evaluation results, the mutual information-based quantification of causal effects yields an impressive 94% accuracy with logistic regression in predicting accident fatalities.