Naslov
Multi-label classification of energy efficiency of public buildings based on random forest and CART

Autori
Has, Adela ; Đurđević Babić, Ivana ; Šebalj, Dario

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
KOI 2022 Book of Abstract / Mijač, Tea ; Šestanović, Tea - Zagreb : Hrvatsko društvo za operacijska istraživanja (CRORS), 2022, 81-81

Skup
19th International Conference on Operational Research KOI 2022

Mjesto i datum
Šibenik, Hrvatska, 28.09.2022. - 30.09.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
random forest, classification and regression trees, multi–label classification, energy efficiency

Sažetak
Non-residential buildings deserve attention when it comes to energy consumption, which should be indicated as an energy class in the energy certificate in Croatia. In this paper, data on 509 public buildings with energy certificates were extracted from the Energy Management Information System. These data included 46 features on meteorological, construction, geospatial and occupational characteristics of the buildings. The objective of this study was to develop and compare a random forest model (RF) and a classification and regression tree model (CART) to classify public buildings into energy classes and to determine the key predictors of each model. These methods were selected for their effectiveness in multiple label classification problems. Both methods used different parameters and hyperparameters to obtain a model with the highest classification accuracy. In this work, the CART method outperformed the random forest method with a classification accuracy of 95.05%. The most important variables in both models were construction characteristics of the building. This study could be useful for policy makers in the field of energy efficiency and energy retrofit.

Izvorni jezik
Engleski

Znanstvena područja
Ekonomija, Informacijske i komunikacijske znanosti

POVEZANOST RADA