Research On Building Energy Consumption Benchmarking Model Based On Regression Analysis And Data Mining

Building Energy Consumption refers to the total commercial energy of a building which is in use.Building Energy Consumption Benchmarking is a method to evaluate the energy consumption of the building by comparing with those of other buildings whose styles and functions are of the same.At present,there are mainly two methods to do benchmarking:the statistical method and the data-mining method.However,there is a disadvantage in common: benchmarking based on a single method does not work out efficiently.In this paper we propose an integrated model of Building Energy Consumption Benchmarking based on "voting",nominated as SAI-Voting model.In this model, three single-models,who are respectively built in accordance with three algorithms, are integrated into SAI-Voting.This paper focuses on the design and establishment of three single-models and the integrated model:(1) Establishing the Stepwise model with stepwise regression method;(2) Proposing a benchmarking model,called Apriori-rule,for the first time utilizing association rule mining technology;(3) Establishing a benchmarking model called ID3 by building a decision tree with ID3 algorithm;(4) Proposing an integrated model of Building Energy Consumption Benchmarking based on "voting",nominated as SAI-Voting model in which the Stepwise model,the Apriori-rule model and the ID3 model are used.The experimental results show that the performance of SAI-Voting model has significant improvement gains over the three models based on single method.Besides, it is shown that the Apriori-rule model is better than the ID3 model in accuracy.