基于用电数据挖掘的老龄独居家庭辨识

Identification of Elderly Living Alone Families Based on Electricity Consumption Data Mining

为老龄独居家庭提供特殊供电服务是电业履行社会职能的重要途径。此类家庭传统上由社区查访获知,工作量大、周期长,电业向社区索取也很困难。为此,在分析独居老人用电特征的基础上,提出一种老龄独居家庭辨识方法。通过选用支持向量机为分类器,并调节样本权重最大化约登指数,解决此类家庭作为小样本的识别问题。案例分析表明,所提方法可获得90%以上的准确率和召回率。研究成果有助于电业推行相关供电服务,也有助于提升此类家庭搜索的覆盖度和及时性。

For electricity supply companies(ESCs),it is an important way of performing social functions to provide the elderly living-alone families(ELAFs)with special services.Traditionally,the ELAFs are found by door-to-door survey of community staffs,which has heavy workload and long cycle problems.It’s also difficult for the ESCs to obtain such information from the communities.For this reason,based on the analysis of the electricity consumption characteristics of the elderly living alone,an identification method for the elderly living alone families was proposed.By selecting support vector machine as the classifier and adjusting the sample weight to maximize the Yoden index,the problem of identifying such families as small samples is solved.The case analysis shows that the proposed method can obtain an accuracy and recall rate of more than 90%.The research results will help the electric industry to implement related power supply services,and also help improve the coverage and timeliness of such home searches.