Outlier Detection in Financial Data Based on Voronoi Diagram

Outliers in financial data can distort computations and give an incorrect picture of the past performance of financial products. The statistical methods used to analyze time series, such as ARMA and ARCH, require special hypotheses, and try to describe the system behavior by using a fixed structure, which is inappropriate to apply to complex financial data, such as high frequency data. This paper introduces a new data mining method to detect outliers in financial data. Based on the Voronoi diagram, we propose a novel method, which called Voronoi based outlier detection (VOD), to provide efficient and effective outlier detection in financial data.