Network User Behavior Analysis Based On Dynamic Association Rule

With the popularity of the Internet and e-commerce development, an increasing number of network data were produced by the massive Internet users’ behavior, such as online shopping behavior, click Web browsing, etc., In this case, research on analyzing the behavior of Internet users during the surfer process has been a hotspot. Between these Web data, some special relationship implied, people are eager to find an effective way and its rules as the decision basis in guiding network operators’ production and operation. By using data mining technology we can dig out the potential regularity of large amounts of data or validate the known rules and knowledge. In general, data mining technology can be divided into the association rule mining, clustering technology, regression analysis and time series model, etc. Association rules mining technology is widely used.In the traditional research, association rules is considered to be a static rule mining based on transaction database data, it doesn’t change over the time factor. However, studies have shown that some actual mining of association rules in database are time characteristics, so we should take the time factor into consideration in the process of data mining and observe the rules change over time. Dynamic association rule mining is based on the interval division, and we evaluate the rules according to the degree of support and confidence level.This paper is based on the relevant theoretical research, according to the online customer transaction data, we build customer behavior analysis model by using dynamic association rule. Large amounts of log data are produced during the process of online transactions of customers. Based on these data, we establish a customer-online shopping goods association model and analyze the customer behavior. Under the premise of information classification of customers and online goods, mining association rules of customer online transaction data by using dynamic association rule mining algorithm. Rules are chosen according to the interestingness, and eliminate the rules which not satisfy the interestingness threshold, then apply gray linear regression to forecast support vectors of the rules, and add the predicted value to the support vectors. Finally, according to the rules and its predicted value, analysis the trend of customers’ online shopping behavior. Verify the model through instance, and then apply the model into the actual online customer data, the application shows that the model has good applicability.