回归法挖掘超期望模式

Mining Against-expectation Patterns Using Regression Analysis

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.