论文部分内容阅读
为探讨习惯性违章行为(HVB)的属性间与属性内的关系特征,建立习惯性违章行为的耦合关联分析模型。首先,分析违章行为属性值的分布特征及关联关系,运用关联规则(ARM)挖掘思想和耦合关系理论对各类违章行为下相应属性的关联系数进行求解,得到一个耦合关联度向量集,且对诸向量从大到小排序;然后,依据排序后耦合关联度向量集映射成习惯性违章行为耦合关联分析模型。最后,引入召回率、精确率和平均绝对值误差(MAE)等3个指标,分别求解数据集和模型的指标结果。所建模型与ARM分析结果的对比表明,模型在习惯性违章行为关联关系分析的准确性与全面性方面都效果良好。
In order to explore the relationship characteristics between attributes and attributes of habitual violation behavior (HVB), a coupled analysis model of habitual violation behavior was established. Firstly, this paper analyzes the distribution characteristics and the relationship of the attribute values of the illegal behaviors, and uses the mining idea of the association rules (ARM) and the theory of the coupling relations to solve the correlation coefficients of the corresponding attributes under various kinds of violation behaviors, and obtains a set of coupled relevance vector Sort the vectors from largest to smallest; and then map the coupled set of correlations into a habitual violation-coupled behavior analysis model. Finally, three indicators, such as recall rate, accuracy and mean absolute value error (MAE), are introduced to solve the data set and model index results respectively. The comparison between the model and the results of ARM analysis shows that the model works well in the accuracy and comprehensiveness of the relationship analysis of habitual and illegal behaviors.