The Analysis Of Customer Churn Prediction Based On Data Mining

With the rapid development of global economy, the increasing competition for customers and expanding market has happened in various industries. Some research results show that the costs to develop a new customer is 3-10 times than to retain an old customer. How to predict,analyze the reasons and formulate relevant strategy of customer churn problem has become a focus in modern enterpriseDue to the rapid expansion of market and the sustainable development of science and technology,the market will produce large amounts of data that has characteristics of imbalance,missing value and noise data stored in the database. Random Forest(RF) can effectively deal with large data,overcome the above shortcomings and achieve good results of regression and classification. Thus this paper will introduce RF in customer churn prediction of a cosmetic brand company.Firstly, this paper use RF to establish customer churn prediction model of a comestic brand company,that mainly contains using existing data to portrait lost customers,using a part of data for training rules and classification, then using the rules to predict the rest of data, and comparing predicted values and real values to calculate the prediction precision of the model. The empirical results show that in the condition of data missing and serious imbalance, RF can forecast lost customersaccurately and achieve a higher prediction precision.After establishing the customer churn prediction model for enterprise data,this paper uses combination of the most common RFM model in CRM and K-means clustering algorithm for customer value segmentation. The combination divide customers into 8 kinds. For each type this paper analysis the characteristics of the three indexes-- R(Recency),F(Frequency) and M(Monetary),and determines the loss of retention value. Finally,this paper calculates the comprehensive value of each customer, orders customers by descending order for each class according to the comprehensive value,makes clear which customer is more worth to retain and provides the specific retention strategy.