Query-adaptive training data recommendation for cross-building predictive modeling

Predictive modeling in buildings is a key task for the optimal management of building energy. Relevant operational data are prerequisite such task, notably when deep learning used. However, not always available, case newly built, renovated, or even yet built buildings. To address this problem, we propose similarity approach to recommend relevant training target solely by using minimal contextual description on it. Contextual descriptions modeled as user queries. We further ensemble most used machine algorithms context predictive modeling. This contributes genericity proposed methodology. Experimental evaluations show that our methodology offers generic cross-building and achieves good generalization performance.