Table 15 Performance comparison of the suggested models with the models from prior studies
Ref | Year | Region/country | Data resolution | Methods | Best performer | Performance |
|---|---|---|---|---|---|---|
Shawon et al. (2021) | 2021 | NA | Hourly | ARMA, ARIMA, SVR, and ANN | Polynomial SVR | RMSE = 0.552 MAPE = 5% |
Mohsin et al. (2021) | 2021 | NA | 3− h interval | BNN, and Lasso | BNN | MAPE = 19.01% NMAE = 0.003 |
Hanoon et al. (2022) | 2022 | 14 regions in Malaysia | Daily | GPR, SVR, and BTs | GPR | RMSE = 0.18144 MSE = 0.03292 NSE = 0.26957 MAE = 0.13498 R2 = 0.38115 |
S. Kumar P (2019) | 2019 | Waterloo, Canada | 15-min interval | BPN, BPN with MIFS, RBF, RBF with MIFS, NARX, and NARX with MIFS | NARX with MIFS | RMSE = 0.5814 MAE = 0.4381 |
Elsaraiti & Merabet (2021) | 2021 | Halifax, Canada | Hourly | ARIMA, and LSTM | LSTM | RMSE = 3.124 MAE = 2.457 |
Liu & Chen (2019) | 2022 | East Jerusalem, Palestine | 3-h interval | MLR, ridge, lasso, RF, SVR, and LSTM | RF | MAE = 0.894 MSE = 1.345 MAD = 0.715 R2 = 0.435 |
Xie et al. (2021) | 2021 | Yanqing, and Zhaitan, Beijing, China | Hourly | ARMA, single-variable LSTM, and MV-LSTM | MV-LSTM | RMSE = 1.1460 MAE = 0.8468 MBE = 0.0276 MAPE = 0.6412 |
Malakout (2023) | 2023 | Turkey | Monthly | LightGBM, GBR, AdaBoost, Elastic net, lasso, and ensemble method (LightGBM and AdaBoost) | Ensemble method | RMSE = 0.2080 MAE = 0.1410 MAPE = 0.0292 R2 = 0.997 |
Krishnaveni et al. (2021) | 2021 | Las Vegas, USA | Hourly | MLR, Lasso, SVR, and MPFFNN | SVR | MSE = 0.011217 MAE = 0.080115 |
This study | 2023 | Kutubdia and Cox's Bazar, Bangladesh | 3-h interval | MLR, Ridge, Lasso, Elastic Net, KNN, DT, RF, GBR, AdaBoost, XGBoost, LightGBM, LSTM and GRU | CatBoost | MSE = 0.3744 MAE = 0.4415 R2 = 0.8670 |