Behavioral Portrait for Electricity Consumers Based on Unsupervised Learning and Feature Engineering
To enable the power supply enterprises to realize accurate classification and characterization of electricity consumers,an electricity consumption feature selection and behavioral portrait method for electricity consumers is pro-posed in this paper.First,a generalized internal evaluation index for clustering is comprehensively utilized to realize the automatic determination of electricity consumer categories.Then,with the consideration of the possible nonlinear re-lationship between feature variables,a heuristic algorithm is used to realize the preferred feature set of electricity con-sumers by integrating the mutual information and distance correlation indexes.Finally,based on the preferred feature set,the radar chart visualization of electricity consumption behavioral characteristics of electricity consumers is carried out,which realizes the behavioral portrait portrayal of electricity consumption behavior and provides a basis for enter-prises to provide targeted services for electricity consumers.
万方数据
维普
