Sleep Predict Models Based On Logistic Regression

There are a lot of potential customers so-called sleepers on the plat form of third-party payment. It regard sleepersas customer churn in the case of producing no profit, which causes big wasting for a company. Hence, for the perspective of company, identify which customers are correct sleepers can effectively find the values of users by the counter measure of marketing department. The research object of paper is to build a model investigating how data from the behavior of users, to know which variables can influence the use of the tool.On univariate analysis, using "elogit" scatter diagram analysis of continuous variable variables transformation and determine the effect of the model. The frequency analysis are used analysis of categorical variable. Use binary logistic model to establish mobile payment platform users sleep prediction model, combined with the actual data for empirical research.Results:after the last transaction balance, recent consumption number in March, frequency of consumption accounted for than in the past three months, in the last three months of transaction value of the total turnover accounted for months, the last three months of transfer number proportion, real-name level0, recently march transfer amount proportion and the last4to6months transfer amount than change percentage of users to sleep with a statistical significance. After the last transaction balance, recent consumption number in March, the last three months of consumption sum of several accounted, transaction amount of months in the past three months the total turnover affect the most significant.