基于异源交叉回归分析的电力客户缴费风险建模

摘要

客户缴费风险预测是电力企业实施个性化服务、提升营销效率的有效举措之一.文中首先提出了电力客户缴费风险的归一化度量方法,并分析了缴费风险影响因素的异源数据特性;然后引入异源交叉的回归方法,对典型电力客户缴费数据进行回归分析.文中通过分析电力客户缴费风险影响因素情况,提出了一种基于异源交叉回归的电力客户缴费风险分析新方法.研究结果表明,电力客户的缴费风险影响因素中,其历史缴费情况的相关性影响较大,其次是其用电量变化,而客户所属行业、地区等因素的总体影响程度较小.%Payment risk predication of electricity consumers is an underlying task to implement personalized custom service.In the study,a novel normalized payment risk measure,a heterogeneous fusion regression model and a heterogeneous cross validation method are respectively put forward.By analyzing the influence factors of the payment risk of electricity consumers,a new method based on heterogeneous cross regression is proposed.The experimental results indicate the effectiveness of our methods,and show that,personal historical payment information is the most crucial influence factor to payment risk,personal historical electricity consumption is the second most important factor,and consumers′ trade and area attributes are slightly related to payment risk value.