具有下界约束的关联规则算法研究

Mining association rules with lower bound constraints

约束关联规则是关联规则研究中的重要问题,目前的研究大多集中在单变量约束,对双变量约束的研究较少,而双变量约束在实际中也有重要作用。针对这种情况,提出了双变量约束中具有下界约束的关联规则问题。在此基础上,给出了下界约束的定义,然后分析了满足下界约束频繁集的性质,并给出了相关的证明。最后提出了基于FP-Tree的下界约束算法,采用了预先测试的方法,降低了需要测试项集的数量和计算成本。实验结果表明,该算法具有较高的效率。

Mining Association rule with constraints is important research area in association rules mining. At present single variable constraints is mainly researched, only a few is on double variable constraints. So mining association rule with lower bound constraints isstudied, and lower bound constraints is one kind ofdouble variable constraints. Based on this the definition oflower bound constraints is presented, then property of frequent item sets which satisfying lower bound constraints is analyzed and proved. And an algorithm of mining association rules with lower bound constraint is presented, which based on FP-Tree. In this algorithm the method of test in advance is adopted, which reduce the cost of item set testing, boosts performance. Experimental test show the algorithm is quite efficient.