客户流失分析外文翻译

客户流失分析外文翻译内容摘要：

lutionary learning Buckinx et al. [4] Retail business 158 884 customers Logistic regression, ARD automatic relevance determination, decision tree Buckinxet al. [5] Daily grocery 878 usable responses MLR multiple linear regression, ARD, and decision tree Ferreiraet al. [10] Wireless tele 100 000 subscribers Neural work, decision tree, hierarchical neurofuzzy systems, rule evolver Gatland[11] 1 Retail banking 100 customers Multiple regression Hwanget al.[14] Wireless tele 16 384 customers Logistic regression, neural work, decision tree Mozeret al. [16] Wireless tele 46 744 subscribers Logistic regression, neural work, decision tree Table 1 presents examples of the churn prediction studies found in literature. The methods used for churn analysis are presented in the table along with a case data size and market sector information. Buckinx et al. measures the loyalty and churn rate differently in retail setting. The loyal customers are those who shop frequently and at the same time exhibit a regular buying pattern [4]. In this retail setting the customer churn is defined as customers who switch their purchases to another store. This is hard to detect because customers may still have transactions in the previous store. So Butnix et al. classify the customer a partial defective if hedeviates from his established buying behavior [4]. This is possible because in their setting they focus only on loyal clients Personal retail banking sector is a typical market sector where a customer is not regularly switching from one pany to another. Customers usually give their banking business to one or two panies for long periods of time. This makes customer churn a priority for most panies in the banking sector. Garland has done research on customer profitability in personal retail banking [11]. Although their main focus is on the customers’ value to the study bank, they also investigate the duration and age of customer relationship based on profitability. His study is based on customer survey by mail which helped him to determine the customer’s share of wallet, satisfaction and loyalty from the qualitative factors. 3. Methods . Logistic regression Binomial binary logistic regression is a form of regression which is used in a situation when dependent is not a continuous variable but a state which may or may not happen, or a category in a specific classification [8]. Logistic regression can be used to predict a discrete oute on the basis of continuous and/or categorical variables. Multinomial logistic regression exists to handle the case of dependents with more classes than two. Although logistic regression has been used in variety of areas, for example in childhood ADHD context [19], logistic regression has also been used in customer analysis. For example Buckinx et al. have used logistic regression for predicting partially defect customers in retail setting [4]. Multinomial regression has been used for predicting the customer’s future profitability, based on his demographic information and buying history in the book club [1]. In the logistic regression there can be only one dependent variable. Logistic regression applies imum likelihood estimation after transforming the dependent into a logistic variable [8]. Unlike the normal regression model the dependent variable in logistic regression is usually dichotomous: the dependent variable can take value 1 with probability q。