基于优势粗糙集的交通事故多属性关联分析

Multi-attribute correlation analysis of traffic accident based on dominance rough set

现有交通事故指标关联分析的粗糙集模型都是基于等价关系且只考虑单个决策属性的情形,并未考虑交通事故指标的多元性及其指标值大小的比较,在实际应用中存在着局限性。文章以交通事故数据为背景,基于隐患指标和事故指标2个一级指标,构造了一种多决策属性优势粗糙集模型,分析了模型的相关性质及其与单决策属性模型的区别与联系,定义了事故指标关于隐患指标的分类质量,基于该分类质量得到了隐患指标相对于事故指标的约简集,根据约简结果可以导出确定性最简序决策规则及可能性最简序决策规则,最后结合具体的交通事故数据验证了模型的有效性。

The existing rough set model for traffic accident index analysis is only suitable for containing a single decision attribute and has not taken into account the diversity of traffic accident index and the comparison of their index values which has limitation in practical applications.Under the background of traffic accident index,a novel multiple decision attribute dominance-based rough set model is constructed based on two first-level indices,namely hidden danger index and accident index.The corresponding properties of model and the difference as well as the relation compared with the single decision attribute model are analyzed.Moreover,the classification quality of the accident index according to the hidden danger index is defined and the relative reduct sets are gotten from which the deterministic and possible order decision rules are derived.Finally,the validity of the proposed model is verified combining with the specific traffic accident data.