粗集理论对股票时间序列的知识发现

提出了将粗集理论应用于时间序列的知识发现。知识发现的过程包括时间序列数据预处理、属性约简和规则抽取三部分。其中数据预处理主要用信号处理技术清洗数据,然后将清洗后的时间序列按照某个变量的变化趋势进行分割,分割后每个时间段内的变化趋势不变,从而将时间序列转换成为一系列静态模式(每种模式代表一种行为趋势),从而去掉其时间依赖性。把决定各种模式的相关属性抽取出来组成一个适用于粗集理论的信息表,然后采用粗集理论对信息表进行属性约简和规则抽取,所得到的规则可以用于预测时间序列在未来的行为。最后将该方法用于股票的趋势预测,取得良好效果。

Knowledge Discovery in Stock Market Time Sieres Based on Rough Set

This paper applies Rough set to the knowledge discovery of time series.The process of knowledge discovery in time series includes preprocessing of time series data,attributes reduction and rules extraction.The preprocessing cleans the raw data using the signal processing techniques.Then,the time series is partitioned to a set of pattern(each pattern represents a trend of time series)according to the trend of certain variable.An information table is formed by the most important predicting attributes and target attribute identified from each pattern.This information table is suitable for the Rough set to discover knowledge.Then we use Rough sets to reduce the attributes and extract rules from information table.The extracted rules can predict the time series behavior in the future.We demonstrate our method on time series of stock market data.