基于概率统计的多维关联数据动态挖掘仿真  

Simulation of Dynamic Mining for Multi-Dimensional Associated Data Based on Probability and Statistics

作　　者：张平[1] 关丽红[1] ZHANG Ping;GUAN Li-hong(School of Science,Changchun University Changchun Jilin 130022,China)

机构地区：[1]长春大学理学院,吉林长春130022

出　　处：《计算机仿真》2022年第3期402-406,共5页Computer Simulation

基　　金：国家自然科学基金青年基金项目(11801041)。

摘　　要：为有效挖掘所需的多维关联数据,及时获取多维关联数据中的可用信息,结合概率统计理念,构建出多维关联数据动态挖掘方法。基于核密度估计的非参数概率密度预估形式,结合实际的数据分布先验知识,设计半参数化概率密度预估模型。根据制定的参数设置原则,合理设置模型参数,利用改进的遗传优化算法,提取出多维关联数据间关联规则。将滑动窗口设定成动态采集窗口,利用界定规则动态采集数据,令连续两个窗口部分重合,反复求解叠加区域的边界数据,实现数据动态处理,依据时间门限值,采用K标号方法动态控制、挖掘目标数据。仿真阶段,以道路交通事故数据为挖掘目标,提取关联规则,经对比不同情况、不同时段的事故数量,验证出上述方法挖掘结果数据关联性高,能够为相关领域提供有效的参考依据。The dynamic mining method of multidimensional association data based on the concept of probability and statistics was established in this paper.In order to effectively mine the required multidimensional association data and obtain the available information in multidimensional association data in time,according to the nonparametric probability density prediction form of kernel density estimation and the prior knowledge of actual data distribution,the semi-parametric probability density prediction model was designed in this work.The parameter setting principle was formulated to set the model parameters.The improved genetic optimization algorithm was used to extract association rules between multidimensional association data.The sliding window was set as a dynamic acquisition window.Defining rules for dynamically collecting data were used to partially coincide two consecutive windows.Meanwhile,the boundary data of the superimposed area were solved repeatedly to dynamically process these data.The k-label method was introduced to dynamically control and mine the target data according to the time threshold.In the simulation stage,taking the road traffic accident data as the mining target,the association rules were extracted.By comparing the number of accidents in different situations and different periods,it was verified that the mining results of the above method have high data relevance and can provide an effective reference basis for relevant fields.

关 键 词：概率统计 多维关联数据 动态挖掘 关联规则 核密度估计 支持度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]