基于AP聚类与随机森林的客户流失预测研究

利率市场化、大数据迅速发展,银行业均表现出明显的“二八定律”现象,20%的优质客户占据了银行的大部分资产。那么,如何防止银行客户流失,尤其是优质客户的流失,已经成为银行越来越关注的问题。因此,建立优质客户流失预警模型就显得尤为重要。以某商业银行为例,重新对客户流失进行定义,重点关注银行优质客户的流失预警,首先使用AP聚类算法进行属性选择,然后使用随机森林方法建立客户流失预警模型,预测零售优质客户未来3个月流失的可能性。为了验证该方法的有效性,首先在UCI数据集上进行验证,得到了较好的效果,然后使用该方法构建银行业优质客户流失预测模型,实验结果表明该模型的实际预测效果相较于一般的决策树方法,具有更高的准确性。

Research on Prediction Model of Customer Churn Based on AP Clustering and Random Forest

With the rapid development of interest rate marketization and big data,the banking industry has shown a clear“the 80/20 Rule”phenomenon.20%of customers occupy most of the bank’s assets.Therefore,how to prevent the loss of bank customers,especially the loss of high-quality customers,has become a growing concern of banks.It is particularly important to establish the model of high quality customer loss warning.Taking a commercial bank as an example,we redefine customer churn according to the actual bank marketing operation and focus on the bank’s high-quality customers.Firstly,AP clustering algorithm is used for attribute selection,and then random forest method is used to establish an early warning model of customer loss,so as to predict the possibility of customer loss of retail high-quality customers in the next three months.In order to verify the effectiveness of the proposed method,it is firstly verified on the UCI data set and a ideal result is obtained.Then,the proposed method is used to structure the prediction model of Bank Customer Churn.The experiment shows that the actual prediction effect of this model is more accurate than that of the general decision tree method.