Table 3.

A summary of AI-Big data analytics frameworks proposed for energy forecasting in buildings

Ref.	AI model	Forecasthorizon	Building nature	Year	Description	Evaluation metrics
						RMSE	MAE	MAPE	Others
Skomski et al. (2020)	seq2seq	Short-term	Office	2020	Demonstrate the efficiency of seq2seq RNNs for load prediction using a restricted feature set	✓			nRMSE
Bessani et al. (2020)	Bayesian networks	Short-term	Residential	2020	Handle the volatility and the uncertainty of buildings’ loads	✓		✓	nRMSE, MedAE
Ribeiro et al. (2018)	transfer—based MLP, SVR	Long-term	Residential	2020	TL-based trend and seasonal adjustments to predict cross-building load		✓		MSE
Ahmad et al. (2020)	GSD-GPRM, RBDT, BBRT, BMCDT	Short-, long-term	Office	2020	Building load prediction in non-climate sensitive and climate-sensitiveconditions		✓	✓	CV
Moon et al. (2018)	ANN, SVR, PCA-FA	Short-term	Academic	2020	Energy prediction of higher educational institutions	✓	✓	✓
Zhang et al. (2020)	LSTM, GRU, CIFG	Short-term	Public	2020	Hybrid DL-based energy prediction combined with an interpretation process	✓	✓		CV-RMSE, $R^2$
Wen et al. (2020)	RNN-GRU	Short-, mid-term	Residential	2020	Achieve well performance with limited input variables	✓	✓	✓
Park et al. (2020)	XGBoost, RF, DNN	Short-term	Industrial	2020	A Two-Stage energy consumption prediction			✓	CVRMSE
Khamma et al. (2020)	GAMs	Short-term	Office	2020	Embed domain knowledge and prior understanding of buildings into the prediction model				CVRMSE, NMBE
Somu et al. (2020)	ISCOA-LSTM	Short- , mid-, long-term	Residential	2020	Accurate and reliabale data driven load forecasting	✓	✓	✓	MSE,Theil U1, U2
Liu et al. (2020)	A3C, DDPG, RDPG	Mid-, long-term	N/A	2020	Improve the forecasting accuracy with increasing computation time	✓	✓		$R^2$, CV
Zhang et al. (2020)	DBN-DEEM	Short-term	Residential	2020	Predict stochastic energy consumption using Cyclic feature (CF) extracted via spectrum analysis	✓	✓	✓	r
Lu et al. (2020)	CEEMDAN-XGBoost	Short-term	Intake towers	2020	Have half of prediction error of XGBoost using real-world data for a period of 8 years				RMSPE, Theil U1, U2
Wang et al. (2020)	stacking model	Short-term	Academic	2020	Building load forecasting using model integration	✓	✓	✓	CVRMSE
Somu et al. (2021)	$\kappa$CNN-LSTM	Long-term	Academic	2021	Capture the load spatio-temporal features and aid in decision making	✓	✓	✓	MSE
Yuan et al. (2020)	WNN-cuckoo search	Mid-term	Commercial	2020	Optimally tuning the WNN parameters CS with				DMAPE, AE
				2020	a real-world validation
Mawson and Hughes (2020)	DFNN, RNN	Mid-term	Industrial	2020	Load forecasting and condition monitoring in manufacturing buildings	✓	✓
Bui et al. (2021)	LSTM	Long-, and short-term	Residential	2021	Multi-behavior with bottleneck features LSTM for to predict energy consumption	✓	✓	✓	NRMSE
Dun and Wu (2020)	Grey model	Long-term	Residential	2020	Load forecasting of three kinds of buildings, i.e. rural, public and urban buildings			✓
Khan et al. (2021)	LSTM-KF	Short-term	Residential	2021	Learning to statistical model for ensemble predicting of energy consumption	✓	✓	✓
Li et al. (2021)	TL-based ANN	Short-term	Residential		Load prediction of information-poor buildings			✓	NTR
Grolinger et al. (2016)	NN-SVR	Short-term	Sport-venues	2016	Load forecasting in a challenging scenario with high variations caused by the hosted events			✓
Pinto et al. (2021)	RF, GBR	Short-term	Office	2021	Combine multiple learners to optimize the learning process			✓