Published Paper Details:

The total number of motor vehicles is still increasing at a high rate due to the social economy's rapid expansion and the speed at which cities are becoming more populated. Large and medium-sized cities' roads are getting more and more congested, which increases the frequency of traffic accidents. Finding accident hotspots early on is crucial to improving road safety and lowering the number of traffic accidents. Eight impact factors (holiday, day of week, time, rush hour traffic, accident location type, accident type, weather, responsibility, and black spot) were set for the analytic dataset in this study, which included data from traffic accidents on the Lianfeng Middle Road, Yinzhou District, Ningbo City. The classic Kmeans clustering algorithm's drawbacks were addressed by the proposal of the enhanced algorithm, which is vulnerable to early clustering centres and outliers. The dataset's traffic accidents were split into two groups using this algorithm: black spots and non-black spots. After that, we used the new dataset to build a black spot recognition model using a Bayesian network, and we compared it with other popular methods including the ID3 decision tree, logistic regression, and support vector machine. As demonstrated by the values of the ROC area, TP rate, FP rate, accuracy, precision, recall, F-measure, and F-measure reaching 0.618, 0.668, 0.580, 0.650, 0.668, 0.590, and 0.668, respectively, the Bayesian network was the most successful model for locating black spots associated with traffic accidents. In addition, a bivariate correlation model was used to confirm the impact factors and black spots' link. The findings showed that Black spots, which had a value of sig<0.05, showed significant associations with the accident location type, accident type, time, and responsibility. In order to greatly improve road safety, the conclusions may offer reference data for the detection and avoidance of high-risk areas for accidents.