Research on personalized referral service and big data mining for e-commerce with machine learning

Abstract

Machine learning (ML) is one of the main methods to address the problem of big data mining. ML can enable the e-commerce system upon self-innovate and improvement by accumulating prior knowledge. The big data produced by transaction, interaction and observation from e-commerce enterprises can greatly provide decision-making service for marketing strategy. In this paper we take the e-commerce data of tea-device enterprise as an example, use the FP-grow algorithm to get the frequent item sets, so as to mining and analyze association rule of user behavior to get feature vector as the basis of user classification, then use Naive Bayesian algorithm on the feature vector to implement clustering learning for precision marketing and personalized online referral services. Finally, we evaluate the feasibility of big data mining with ML through the profit produced by the sale of goods. Experimental results show that ML can not only greatly improve the performance in big data mining, can also achieve precise marketing, and can further increasing about 20% marginal profit for each type of goods.