Table 1.

The performance of the previous studies on the prediction of MW generation

Author(s)	Province(s)/Country	Methods	Best Model Performance (R, R2)
Bdour et al. [9]	Irbid, Jordan	MLR	R2 = 0.918
Sabour et al. [10]	Gilan, Iran	Linear Regression	Not available
Jahandideh et al. [11]	Fars, Iran	MLR, ANN	R2 = 0.990 (ANN)
Eleyan et al. [12]	Jenin, Palestine	System Dynamics Models	Not available
Idowu et al. [13]	Lagos, Nigeria	MLR Models	R2 = 0.998
Karpušenkaite et al. [14]	Lithuania	ANN, MLR, SVM, and different non-parametric regression methods	R2 = 0.905 (with GANPR using regional dataset) R2 = 0.986 (with SSNPR using long annual dataset)
Tesfahun et al. [15]	Ethiopian	Mathematical predictive models from the available literatures.	R2 = 0.965 (with the number of inpatients) R2 = 0.424 (with the number of outpatients)
Al-Khatib et al. [16]	Nablus, Palestine	MLR	R2 = 0.984
Chauhan and Singh [17]	Uttarakhand, India	ARIMA models	R2 = 0.832 with ARMA(1,1) model
Minoglou and Komilis [18]	41 Countries	MLR and Principal Component Analysis (PCA)	R2 = 0.8473
Adamović et al. [19]	European countries	General Regression Neural Network (GRNNs) Models	R2 = 0.999 (for the prediction of chemical hazardous waste) R2 = 0.975 (for healthcare and biological hazardous waste)
Karpušenkaitė et al. [20]	Lithuania	Time Series Moving Average, Time Series Holt’s Exponential Smoothing, Hybrid Model	Not available
Thakur and Ramesh [21]	Uttarakhand, India	MLR, ANN, and Polynomial Regression	R2 = 0.954 (ANN for total waste)
Golbaz et al. [22]	Karaj, Iran	MLR and several Neuron and Kernel based machine earning methods	R2 = 0.82 – 0.86 (Kernel-based models) R2 = 0.68 – 0.74 (Neuron-based models)
Hao et al. [23]	Shanghai, China	GM (1,1), Triple Exponential Smoothing (TES), Particle Swarm Optimization (PSO) Optimized Back Propagation (BP) Neural Network, and Hybrid Model	Not available
Çetinkaya et al. [24]	Aksaray, Turkey	MLR	R2 = 0.979

*SSNPR: Smoothing splines non-parametric regression

*GANPR: Generalized additive non-parametric regression