Page 86 - New Trends in Green Construction

Energy Efficient Building
integrated approach to make predictions and also allows conducting sensitivity analysis.
3. Results and conclusions
Fifteen INLA models were run, firstly combining the response variable with each covariate separately, and finally, another model integrating all covariates. The results showed that the inclusion of a higher number of covariates improves the model since DIC (deviance information criterion) and CPO (conditional predictive ordinate) become lower, the correlation coefficient between the predicted and observed values increase and the RMSE (root mean square error) decrease. The results of the best fit model are presented in Figure 2.
DIC: 1224.61 CPO: 56.4266 Correlation: 0.921655 RMSE: 1394.517
Figure 2. Results of the best fit model including all covariates
The INLA statistical analysis enabled us identifying that the most significant covariate on the final energy consumption is the number of dwelling’s occupants, which demonstrates that usage habits play an important role on building energy performance assessments.
4. Acknowledgment
The authors are grateful to Universitat Jaume I, Convocatòria d'ajudes postdoctorals per a la incorporació a grups d’investigació de l’UJI (Spain). (Contract POSDOC-A/2017/17).
5. References
Braulio-Gonzalo, M., Juan, P., Bovea, M.D., Ruá, M.J., 2016. Modelling energy efficiency performance of residential building stocks based on Bayesian statistical inference. Environ. Model. Softw. 83, 198–211.
CTE, 2013. Código Técnico de la Edificación. Spain.
Kavgic, M., Mavrogianni, a., Mumovic, D., Summerfield, a., Stevanovic, Z., Djurovic-Petrovic, M., 2010. A
review of bottom-up building stock models for energy consumption in the residential sector. Build. Environ.
45, 1683–1697.
Rue, H., Martino, S., Chopin, N., 2009. Approximate Bayesian inference for latent Gaussian models by using
integrated nested Laplace approximations. J. R. Stat. Soc. 72, 319–392.
86