基于Logistic回归模型的电力客户欠费违约概率的预测

在分析了引起电力客户欠费原因的基础上，利用可以获得的数据设计了欠费风险识别模型中的关键影响变量，并运用Logistic回归理论与方法建立了可识别电力客户欠费可能性大小的模型，根据所掌握的客户最新资料可提前预测出其欠费的违约概率，从而改变了原来的事后欠费管理的被动局面，达到了减少电力客户欠费风险的目的。

Predication of Default Probability of Clients' Electricity Charges Arrears Based on Logistic Regression Model

On the basis of analyzing the reason bringing on electricity charge arrears and utilizing the gettable data, the key variables impacting the charge arrears, which are needed in arrear risk recognition model, are designed; by means of logistic regression theory and method that can deal with bi-category problem, a model that can recognize the possibility extent of electric power clients’ charge arrears is established. Thereby, it is possible to predict the default probability according to the latest client data that the electric power enterprises could acquire, and the previous post-mortem management of charge arrears can be turned into a priori one, therefore the propose of preventing client arrears could be achieved.