零售商品关联大数据稀疏网络的快速聚类算法

关联规则方法被广泛应用于分析零售企业交易数据,以此指导品类管理、门店布局陈列和商品促销等运营决策,但面对电子商务网站非常巨大的数据量,仍存在效率低下的问题.对此,提出商品关联大数据稀疏网络快速聚类算法.首先,利用单步链表结构存储零售商品的共同购买关系矩阵;其次,对商品关联大数据稀疏网络的低度商品节点进行剪枝,降低搜索空间;再次,利用模糊k均值聚类对商品关联大数据稀疏网络进行快速聚类,并利用高连接度值商品节点被低连接度值商品节点分割的思想对剩余节点聚类;最后,将所提算法应用到亚马逊网站商品交易数据分析中,取得了良好的效果.

Fast clustering for sparse network of retail products associated big data

The association rules method is widely used in the analysis of retail trading data so as to guide the operational decision-making for category management, store layout and the commodity sales promotion of products. However, the data of the electronic commerce website is very huge, which leads to the low inefficiency of the associate rule. The fast clustering for sparse network of retail products associated big data is proposed. Firstly, the structure of the one step linked list is used to store the co-purchasing matrix. Then, the nodes with the low degree in the sparse network of retail products associated big data are pruned, which reduces the search space. Furthermore, the nodes of the sparse network of retail products associated big data are grouped by fuzzy k clustering with the idea that the nodes with high connectivity value are partitioned by the nodes with low connectivity value. Finally, the proposed algorithm is applied to analyze the trading data of the Amazon website, and the results show the effectiveness of the proposed method.