摘要
在交易数据库的挖掘研究中,模式未来行为的预测已成为用户关注的焦点。通常是通过分析和挖掘历史交易数据库中的数据得到模式未来行为——频繁出现程度,以便帮助用户决策。为解决这类问题,设计一个基于回归法的算法挖掘一类新模型超期望模式。实验结果表明这种模式在模式评价和模式预测方面很有效,它同时为减少冗余规则提供了一种可行的方法。
In transaction database mining ,predicting the future behavior of patterns has become an important research topic. Predicting the future behavior of patterns usually depends upon identifying the frequency degree of patterns by analyzing and mining the data in historical transaction databases. To address this issue,this paper designs a regression-based algorithm for mining a kind of new pattern. Against-expectation patterns. Authors experimentally evaluate their approach ,and demonstrate that it assists in the pattern evaluation and pattern forecast,as well as provides a feasible method that can decrease the redundancy rules.
出处
《广西师范大学学报（自然科学版）》
CAS
北大核心
2006年第4期79-82,共4页
Journal of Guangxi Normal University:Natural Science Edition
基金
澳大利亚创新基金资助项目(ARC:DP0559536)
国家自然科学基金资助项目(60463003)
关键词
模式
超期望模式
频繁模式
pattern
against-expectation pattern
frequent pattern
