摘要
本文选取2022年抽样的淘宝数据进行分析与挖掘,并基于K-means算法对买家进行聚类分析,初步筛选出疑似刷单行为的买家和卖家。在剔除这些用户后,又利用回归分析法分析卖家获得评价、信用评价体系、卖家店铺等级对销量的影响;采用LSTM算法对销量数据的时间序列进行预测;通过Apriori关联规则算法找到买家与卖家和商品之间的关联。其中,在卖家获得评价对销量的影响中,建立奖励函数来描述好评和差评的影响,结果显示奖励函数与销量呈正相关关系。在信用评价体系对销量的影响中,服务和发货对销量的影响较大。卖家店铺等级,则无明显关系。预测的销量数据虽没有较好的结果,但给出了合理的解释。关联结果显示,买家与卖家和商品之间有一定的联系,本研究仅供参考。
This article analyzes and explores Taobao data sampled in 2022,using the K-means algorithm to conduct cluster analysis of buyers and preliminary screening of suspected brush order behavior among buyers and sellers.After removing these users,regression analysis is employed to analyze the impact of seller evaluations,credit rating systems,and seller store grades on sales.The LSTM algorithm is utilized to forecast sales data time series,and the Apriori association rule algorithm is applied tofind associations between buyers,sellers,and products.In the analysis of the impact of seller evaluations on sales,a reward function is established to describe the influence of positive and negative feedback,showing a positive correlation between the reward function and sales.In terms of the impact of the credit rating system on sales,service and delivery have a significant impact on sales,while there is no apparent relationship with the seller's store grade.Although the predicted sales data does not yield satisfactory results,reasonable explanations are provided.The association results show that there is a certain connection between buyers,sellers,and products.This research is for reference only.
作者
徐晨旸
Xu Chenyang(China Jijiang University,Hangzhou,Zhejiang 310000)
出处
《中国商论》
2024年第7期82-85,共4页
China Journal of Commerce
关键词
数字贸易
数据挖掘
聚类分析
回归分析
时间序列
关联规则
digital trade
data mining
cluster analysis
regression analysis
time series
association rules
