Design And Implementation Of Customer Value Clustering And Churn Prediction System

Since the 21 st century,all kinds of business developing rapidly,which brings the enormous increase of customers for the enterprise to retain,but at the same time an unprecedented number of clients also brought difficulties for the enterprise,how to distinguish between different customer groups,for different customer groups how to make a plan of service and how to accurately predict before customer churn,this several problems which anyway is inevitable for enterprises,this paper is under this premise to discuss how to use the machine learning and deep learning knowledge to solve the problem of this type.When customers are large,we need to differentiate customers from customers to better carry out accurate sales and promotion.The data is based on real data,and there will be errors and accidental factors,so data cleaning and preprocessing are required prior to major steps.Because there is no customer group labels in advance,so when dealing with the problem using clustering algorithm,combined with the most widely applied in the economics of RFM method of value analysis theory through customer features as a clustering algorithm main input.Then using analytical means such as radar map plot different customer groups to data analysis and data visualization,.Finally use deep learning of neural network to predict customer churn,given the loss of customer list for enterprises to carry out specific operations.The whole project adopts the process method,using python2.7 as the development environment,scikit-learn is the development framework.Through the use of machine learning and deep learning techniques we can mining the potential customers and customer churn,avoid the traditional phone hair information waste of resources this low feedback sales behavior,greatly saves manpower and material resources of enterprise's late input,makes the customer management of the enterprise is more efficient and agile.