基于Hash的Gibbs采样关联规则挖掘快速算法

A Fast Algorithm for Gibbs Sampling Association Rule Based on Hash

Gibbs采样算法作为一种特征选择方法在关联规则挖掘方面具有较好的效果。为了进一步加快基于Gibbs采样的关联规则挖掘算法,提出了一种基于Hash结构的Gibbs采样关联规则挖掘算法。该算法改进了二分类数据的存储方式,建立十进制数组,将非重复数据及其支持度插入Hash表中。真实数据集实验结果显示,在时间成本上,提出算法与原算法相比有较好的优势,增加的空间成本在接受范围内。

The Gibbs sampling algorithm,as a feature selection method,has shown good performance in association rule mining.In order to further accelerate the association rule mining algorithm based on Gibbs sampling,a Gibbs sampling association rule mining algorithm based on hash structure is proposed.This algorithm improves the storage method of binary data,establishes a decimal array,and inserts non duplicate data and its support into the hash table.The experimental results on real datasets show that the proposed algorithm has better advantages in terms of time cost compared to the original algorithm,and the increased spatial cost is within an acceptable range.