Outlier mining on multiple time series data in stock market

Abstract

With the dramatic increase of stock market data, traditional outlier mining technologies have shown their limitations in efficiency and precision. In this paper, an outlier mining model on stock market data is proposed, which aims to detect the anomalies from multiple complex stock market data. This model is able to improve the precision of outlier mining on individual time series. The experiments on real-world stock market data show that the proposed outlier mining model is effective and outperforms traditional technologies. © 2008 Springer Berlin Heidelberg.