海量运动数据中的差异参数优化挖掘方法  

Quality Parameters of Mass Movement Data Mining Method

作　　者：宁光芳[1] 齐小文[2] 

机构地区：[1]郑州大学体育学院,河南郑州450044 [2]中州大学信息管理中心,河南郑州450044

出　　处：《计算机仿真》2014年第6期398-401,共4页Computer Simulation

摘　　要：研究在海量运动数据中差异参数的准确提取问题。由于在海量运动数据中,连续运动帧之间受到采集时间短、运动幅度变化过小的影响,运动特征数据之间的差异很小。传统数据挖掘方法中对细微差异数据无法完成准确挖掘。提出利用决策树算法的海量运动数据中优质参数挖掘方法。计算不同属性运动数据的支持度,获取对应数据的置信度和所占比率,计算上述运动数据之间的关联性。采用贪心算法构造决策树,以自顶向下递归的方式构造判定树。将运动数据作为决策树中的树叶节点。实验结果表明,利用改进算法进行海量运动数据中优质参数提取,可以有效提高挖掘的效率和精度,去除运动数据中的冗余数据。The extraction method of different parameters in mass movement data was studied in this paper. In the mass movement data, continuous motion frame is affected by short acquisition time and small motion amplitude change, which makes small difference in motion characteristics of the data. This paper presented a data mining method of high quality parameters in the mass movement data based on decision tree algorithm. The support degree of different attribute motion data was calculated, confidence coefficient and rate of corresponding data were obtained, and relevance of the motion data was calculated. The greedy algorithm was used to construct the decision tree, and to construct the decision tree with top - down way recursive. The motion data were used as leaf node in a decision tree. The experimental resuhs show that the use of modified algorithm to extract quality parameters in the mass movement data can effectively improve the precision and efficiency of mining, and remove the redundant data in the motion data.

关 键 词：运动数据 优质参数 数据挖掘 决策树 

分 类 号：F127[经济管理—世界经济]