Self-organization map method for pattern matching of stock time series

被引：0

作者：

Guo, Wei ^{[1
]}

He, Pilian ^{[1
]}

Wang, Zhong ^{[1
]}

机构：

[1] Tianjin Univ, Dept Comp Sci, Tianjin 300072, Peoples R China

来源：

Proceedings of 2006 International Conference on Artificial Intelligence: 50 YEARS' ACHIEVEMENTS, FUTURE DIRECTIONS AND SOCIAL IMPACTS | 2006年

关键词：

self-organization map; time series; patterns stock;

D O I：

暂无

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

It is difficult to forecast the tendency for stock market since it is a typical non-linear dynamic system. With proper experiences in share market, people can know more about stock time series patterns. Consequently they can determine how and when to invest by use of these patterns. But it is difficult for them to handle such large data set. This paper presents a method of patterns mining for stock time series by use of Self-organization map. Because of considerations in investment income and risk control, there are. 30 dimensions for inputs and 50 for outputs. This method will be helpful in improving stock patterns mining for future investment reference.

引用

页码：118 / 121

页数：4