关键词: 数据挖掘, 时间序列, 证券预测, 相似性搜索, 分段处理

Abstract:

For better analysis of stock exchange data, we give a brief introduction of the generation and application of time series in data mining and by comparison and analysis of different ways of time series processing and relevance searching, we choose a reasonable method to design a small system which is used for stock forcasting. First, the time series are processed properly and the similarity search is preformed and then analyse whether the shortterm trend is the resurgence of history. The final result is anlysised. Experiments showed that the method can find the similarity trends of the stock exchange data in history, and by analyzing similarity trends in history, current trend can be predicted.

Key words: data mining, time series, security analysis, similarity search, segment