WeightLOFCC: A Heuristic Weight-Setting Strategy of LOF Applied to Outlier Detection in Time Series Data

摘要：

Finding outliers is more interesting than finding common patterns in many KDD applications. The local outlier factor method (LOF) is a popular approach to detect outliers, in which a degree of being an outlier will be assigned to each object. In this paper, we present a modification method called WeightLOFCC to better handle outliers in time series data. Differing from the traditional LOF algorithm, the proposed WeightLOFCC method utilizes the idea of semi-supervised learning and weight factor to model data, and makes use of the cross correlation to measure the similarity. We evaluated the proposed algorithm on a large variety of data sets, and the experiment results show that for most of the data sets, our solution for outlier detection can achieve the best performance compared with other classical techniques.

关键词： WeightLOFCC time series data cross correlation weight method

作者: Hongrui Xie Yujiu Yang Wenhuang Liu

作者单位: Information Science and Technology Division,Graduate School at Shenzhen Tsinghua University,Shenzhen 518055 P.R. China

会议类型: 国际会议

会议名称: 6th International Conference on Advanced Data Mining and Applications(第六届先进数据挖掘及应用国际会议 ADMA 2010)

会议地点: 重庆

会议语种:英文

页码: 529-536

在线出版日期: 2010-11-19（万方平台首次上网日期，不代表论文的发表时间）