Modification of predicting parameters of occupants window opening behaviour in residential buildings using Machine learning algorithm

Abstract

In this study, based on the analysis of environmental variables used in the predictive model, we intend to improve the predictive model by reflecting the influence of occupants' propensit y or the measured physical information of the housing unit. Substitutable variables were select ed, and differences in the changes in the importance of environmental variables according to s easons were compared to reflect the variation of occupants. Based on the previous two results, we improved the prediction model. In order to compare the improved effect, the model was c ompared and verified before and after improvement based on the reproducibility, which mean s the window opening prediction. As a result, there is no significant difference in the accuracy of the predictive model, but it can be confirmed that the improvement was appropriate becaus e the recall increased.