Application Of Recommendation Algorithm Combining Two-step Clustering And Association Rules In Pharmaceutical E-commerce

With the advent of the Internet era,e-commerce has developed rapidly,and the products on the e-commerce platform have become more and more abundant,from the daily rise of daily necessities,food,to fresh fruits,medical products,insurance,etc.For consumers,how to choose their favorite products among tens of thousands of products?For merchants,how to recommend their products to consumers?In order to solve these problems,the personalized recommendation system was born,and with continuous research and development,the accuracy of the personalized recommendation algorithm has been greatly improved.This article mainly studies the recommendation algorithm based on association rules.To improve the effectiveness of association rule recommendation,it solves the problem from two aspects:first,before using association rules mining,the data is preprocessed with two-step clustering.The sparse nature of the project leads to the problem that the item cannot appear in the association relationship of the association rule.Two-step clustering is used to group users with similar consumption behaviors into one category,which can improve the problem of sparse user behavior data and improve the appearance of items in the association relationship.Possibility;Second,the adjusted weighted Apriori algorithm is proposed,which weights the items according to the importance of the items,and adjusts the weighted support and weighted confidence calculation formulas based on the New_Apriori algorithm,which improves the weighting in the New_Apriori algorithm The problem of weight cancellation in confidence calculation.This paper combines the data of the medical e-commerce platform to achieve personalized recommendations combining two-step clustering and association rules.Through clustering analysis,users with similar consumption behavior are grouped into one class,and the association rule algorithm is used to mine all types of users purchased in the same category.Relevance between medicine,with strong relevance as a guide to recommend medical products that users may need.Finally,through two groups of comparative analysis,the first group compares the performance of the recommendation system using only association rules and the recommendation system based on the combination of two-step clustering and association rules,and the second group compares the performance of recommendation systems that use only adjusted weighted association rules and those based on a combination of two-step clustering and adjusted weighted association rules.It aims to analyze the performance of the recommendation system combining two-step clustering and association rules and the recommendation performance of adjusted weighted association rules.