Power user portrait model based on random forest

With the vigorous development of energy Internet, all kinds user information data are increasing day by day. How to comprehensively and deeply mine effective users, develop a model predict behaviour characteristics big distinguish customer relationships, provide an accurate basis for next users various platforms have become one research hotspots analysis behaviour. The is sampled according feature vector power user. portrait mining conducted, screening conducted using measure decision tree node purity in model. variable up–down stopping rule generated. Then results Logistics tested analysed, which can effectively proposed strategy based on consumption analysed verify effectiveness scheme. example shows that has strong ability good stability than traditional model, user's advance, reduce complaints, help enterprises form long-term mechanism mutual benefit reciprocity, practical significance. This paper analyses panorama through technology proposes maturity evaluate priority users' electricity consumption. It emphasises use resources methods provided package solve problems consumption, helps companies avoid market risks improve service levels,