Algorithm Of Stock Market Anomalies Data Mining

With the development of the stock market, the stock time series data are increasingly rich, which will urgently need to combine computer information technology to analze and process stocks data, dig out the objective laws of the information hidden in the data. These have very important significance. Time series data mining is to solve this problem and a new type of the emergence data analysis techniques. Using time series data mining techniques to obtain useful information contained in data related to time for achieving the discovery of knowledge and extraction of rules.First, this paper proposes research questions from the practical and theoretical perspective. On reviewing the research status of abnormal data mining techniques and the application, this paper proposes that China’s stock market need further analysis of their data as necessary. This paper has described the practical and theoretical significance of the research, and formed the research content and the basic framework of this article.Secondly, this paper introduces the theory of causes of the stock market anomalies and research methods. China’s stock market’s imperfect operation of its own and policies and systems’s imperfect are the main reasons for the abnormal stock market data generated. By analyzing the basic principles of the existing anomalies data mining algorithms as well as the advantages and disadvantages, found that due to the special nature of stock market data, these algorithms is limited for anomalies data mining on the stock market. On this basis, this paper has defined the anomalies data on stock market, and develops criteria of anomaly detection.Then, this paper proposes a stock market anomaly data mining algorithms. When diging out implicit information on stock time series data, what concerned with is abnormal fluctuations over a period of time, this needs to capture more patterns anomalies and suspected abnormalities. After defining the concepts of time series, this paper has proposed a anomaly generated data dig algorithms based on combination of distance and density (GDD). Through simulation experiments, the results show that the time series anomaly detection algorithm proposed in this paper can effectively detect abnormal fluctuations in the time series.Finally, The GDD algorithm has been applied to the stock market anomalies data mining. Through the statistical description on stock market data, using fractal techniques to determine the size of the sliding window, and then dig anomalies sequence of volatility time series of stock market. Found that the algorithm can effectively dig out abnormal patterns and comparative analysis of the experiment, the algorithm has certain advantages.In short, Time series data mining theoretical research and its effective application in the stock market needs in-depth study, to further improve the efficiency of time series mining tasks and precision requires further exploration.