This research presents an innovative approach to modeling and predicting the heating and
cooling demands of buildings using a simple and fast data-driven model. The research is set
within the context of Model Predictive Control (MPC) systems, which optimize building energy
use by predicting variables and adjusting control inputs accordingly, embodying the principles of industrial ecology by integrating technical, economic, and social dimensions of sustainability; enhancing energy efficiency, reducing CO2 emissions, and improving occupant comfort, which are key goals of industrial ecology. From the available modelling methods:
white, black, and grey box models, to model and predict the heating and cooling demand, a grey box model: multivariate linear regression model with as input actual data selected based on the thermal energy balance and Pearson correlation coefficients. A case study on two rooms: an office and a classroom in the Haagse Hogeschool in Delft serves as the practical application of the developed model. Several models are developed; static and dynamic models, and using different independent variables; indoor surface temperature, outdoor temperature, indoor air temperature, internal heat gains, wind speed and solar light intensity. An accuracy, expressed in the R2 value between 22.52% and 78.57% is achieved with modelling the heating and cooling demand. The developed models are not able to predict the heating and cooling demand, due to multicollinearity between independent variables, overfitting and endogeneity.