A Multi-Model of Classification for Electric-Power Industrial Customer Based on Big Data

Abstract

Aiming to help Electric-Power Industry to fast recognize customer's features, a multi-model of customer classification is proposed in this paper. On basis of big data presented by CCF competition sponsor in China, with some excellent technology or algorithm such as JieBa, SFFS and so on, we extract many important features and successfully draw a portrait for customers who pay close attention to electricity charges. Furthermore, machine learning algorithm and its the strategy selection model are investigated. Eventually, a multi-model is achieved, which can effective draw customer portrait and recognize targeted customer from big data. The result of experiment indicate that the multi-model has precision of 84%, good recall and excellent generalization.