The Analysis And Application Of Data Mining In Electronic Commerce Sales Data

With the speeding up of the modern pace of life in our country, online shoppinghas various advantages in the time, regional, commodities choice and so on then muchcustomers are spring up. With the expansion scale of China network shopping marketand the bulge of netizens, the electronic commerce workbench generates a rich varietyof e-commerce sales data. Facing the complexity of the sales data, the businessmenneed further analysis for the data to dig out some useful knowledge about users’preferences, interesting or buying habits and so on. Based on the knowledge,businessmen can make various sales strategies to meet the users’ demands and finallyobtain more customer resources. Therefore, this paper turns to the sales data ine-commence and applies data mining technology into the electronic business field.Statistical analysis and multi-relational clustering method are used to process the datafrom many perspectives. Through the processing and analysis for the data, we will havea brand new understanding to the influencing factors of users buying behavior, buyinghabits and groups buying characteristics. So that, service providers can use theseknowledge for innovation to the electronic commerce platform, obtaining customersatisfaction, wining the customers resources and gaining more profit. According to thecharacteristics of the electronic business sales data, this work in this paper mainly focuson the following several aspects:1) In order to analysis behavior factors for new commodity buyers in thee-commerce platform, this paper introduced linear regression model in statisticalmethod to study on many factors that influence the new buyers buy different grades ofnew products. Through the feature extraction and the construction of empirical model,some practical statistical results are got from the model. Then the buying behavior ofnew commodity buyers can be understand, different sales strategy can be made for thenew buyers who buy different grades of new products according to different levels ofeach influence factor.2) In view of the dynamic characteristics in electronic commerce sales data, this paper analyzed the dynamic characteristics of commodity sales data based on thecomplex network. The visibility algorithm is used for the data mapping of sales data tothe complex network, through the analysis of the network statistics characteristics tounderstand the fluctuation change of sales. Making use of dynamic wave characteristicsin commodity sales can do deep level analysis for the sales of merchandise andcustomer purchasing information. Thus indirectly to understand the buying habits ofdifferent types of users when buy different commodities. What’s more, this paper madecompared analysis for the sales data in e-commence platform and store supermarket,and then customers’ different purchasing behavior under different channels is analyzedin comparative.3) To make full use of multi-relational information and customer behaviorinformation in the e-commence sales data; to solve the multi-relational clusteringproblem in e-commence field, an effective similarity measurement which combines ofattributes similarity and structural similarity is put out in this paper. The experimentshows that the superiority of this method is verified when compared with othersimilarity measurements. Then, the new similarity measurement is applied intocustomer partition problem. Through the qualitative and quantitative analysis for theexperiment results, more in-depth understanding about the customer behavior obtainedfrom the angle of groups. Moreover, the commodities recommendation experimentverified the new similarity calculation method in multi-relational data.Through the researches on the above problems, the work in this paper can be usedas a system solution for electronic commerce sales data based on the overall analysisand processing. The work of this paper could be used in electronic business field,providing strong technical support and theoretical basis for the personalizedrecommendation technology and marketing strategy. At the same time, this researchmethod can also provide research thinking for other applied problems in some relatedsuitable applicable fields.