The energy efficiency prediction method based on Gradient Boosting Regression Tree

The prediction of building energy efficiency is the key technology in building integrated energy management services. The results of energy efficiency prediction directly reflect the changes of energy efficiency of buildings under the current management level and management strategy. Data analysis can provide guidance for improving management level and adjusting management strategy. Firstly, according to the characteristics of building energy efficiency, an algorithm model of gradient-lifting regression tree is constructed; then, a method of building energy efficiency prediction based on Gradient Boosting Regression Tree model is designed, which is based on a large number of historical data, and established through feature extraction and iterative learning. To satisfy the operational requirements of the model, the first and two derivative of the customized objective function are defined. Finally, the predicted results are explained and analyzed by using LIME with an example, which provides a basis for the formulation of alternative strategy and the improvement of energy efficiency management level.