Research On Product Arrangement Of Online Store Based On Rough Set And Improved Clustering

With the development of the Internet,people's lives got to be more and more dependent on it.Meanwhile,People's basic necessities of life have been closely associated with the Internet.As all we know,the Internet has provided us with a large number of products.How to choose what you need from them becomes a difficult task.The emergence of the recommendation system has greatly reduced the workload of the customer,and also increased the sales volume of online stores.The commodity recommendation system is generally viewed from the user's point,through the user's browsing history or the commodity to recommend products that are more satisfying to the users,however,it is rarely analyzed from the perspective of the network operators and recommend the product from the direction of commodity arrangement.Different from the existing recommendation system which is viewed from the user's point,this thesis puts forward the analysis of sales data from the perspective of the network operator to forecasting the profit of commodity sales,in order to make the arrangement and recommendation of products,so as to help network operators get as much profit as possible.So far,no one else has done similar research.This thesis presents a method of arranging the online store product based on rough set and improved clustering.Firstly,rough set theory is introduced,and the QuickReduct reduction algorithm is used to reduce the conditional attributes that affect the profit of the products.Then,we can get the attributes and its related importance after reduction;Secondly,K-means clustering of data using the weighted Euclidean distance based on attribute importance;Then,the BP neural network algorithm is applied to establish the relevant prediction models of each cluster and predict the profits of various commodities;Finally,arrange and recommend products according to the forecast profitsThe main work of this thesis includes the following sections:(1)Expounding the current situation of online shopping recommendation system,introducing the related research of network information ranking,analyzing and put forward the method of commodity analysis from the perspective of commodity arrangement.(2)Proposing the use of QuickReduct algorithm to reduce various factors affecting the profit of products,and then use the K-means method based on weighted Euclidean distance to cluster the data after reduction,so as to implement the idea of commodity classification.(3)Using the BP neural network to establish profit forecasting model of each cluster,and the commodities are arranged and recommended through the prediction of profits.This thesis takes a supermarket's supply and marketing data as the data sets for experimentation,and through different data processing methods to compare the experimental results,verifying the effectiveness of the proposed method.Experimental results show that the method proposed in this thesis has certain reference value.