Customer Churn Prediction On Website

Internet makes the traditional market a new life, where customer relationship management has entered a new stage. The customer management for enhancing customer loyalty, reducing customer churn is rising. Efficient customer management can help enterprises improve competition to gain advantage in the fierce market. Due to the specialty of the Internet environment, the classical relational contract is no longer applicable. Unstable state makes customer behavior prediction great difficult.This paper analyzed the data and predicts the future purchase behavior of customer by using the theory of the history customer relationship management and data mining. Based on the foreign CD retailer CDNOW customer purchase data set as the research object, we establish the customer relationship Pareto/NBD model, BG/NBD model, the SVM model and LR classifier model, compare the accuracy and the quality of these models, and provide some reference for the analysis of customer purchase prediction in e-commerce website.