基于加速度趋势比较的异常数据清洗研究  

ON OUTLIERS CLEANSING BASED ON ACCELERATING TREND COMPARISON

作　　者：王和勇[1] 吕威[2] 陈冰川[2] 

机构地区：[1]华南理工大学电子商务系,广东广州510006 [2]中山大学软件研究所,广东广州510275

出　　处：《计算机应用与软件》2010年第8期269-270,275,共3页Computer Applications and Software

摘　　要：针对数据集成过程中存在异常数据的问题,提出了加速度趋势比较清洗方法,即比较数据集成过程中某些字段的变化趋势来发现数据的异常。通过加速度趋势比较清洗,提高了数据集成的异常记录检测,改变了数据集成过程中的数据质量,增强了数据的可利用性。最后给出了具体算法的描述。In light to the issue of outliers in data integration process ,we put forward the accelerating trend comparison cleansing method in this paper, namely it finds the outlicrs by comparing the trend of changes in a certain field during the data integration process. The accelerating trend comparison cleansing meliorates the detection of outliers＇ record in data integration, and improves data quality when data is integrating so that the reusability of the data is enhanced. At the end of this article, it gives a specific description of the algorithm details.

关 键 词：数据清洗 趋势比较 异常数据 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]