This version is not peer-reviewed.
Submitted:
09 July 2023
Posted:
12 July 2023
You are already at the latest version
A peer-reviewed article of this preprint also exists.
Methods | MAPE | MAE |
FNN | 2.47 | 163.9 |
GRU | 2.58 | 169.5 |
LSTM | 3.37 | 228.18 |
Naive | 4.93 | 312.34 |
Model | nnodes/layer | nlayers | dropout | epoch | Min MAE | Test MAE | Test MAPE(%) |
FNN | 64 | 2 | 0 | 161 | 214.0 | 212.8 | 3.15 |
GRU | 64 | 2 | 0 | 56 | 243.7 | 210.3 | 2.44 |
LSTM | 64 | 2 | 0 | 200 | 252.2 | 246.2 | 3.86 |
Naive | - | - | - | - | - | 669.0 | - |
Parameters | FNN Results | GRU Results | LSTM Results | ||||||
---|---|---|---|---|---|---|---|---|---|
nnodes | nlayers | look back | dropout | MAE | epochs | MAE | epochs | MAE | epochs |
32 | 1 | 5 | 0 | 226.09 | 319 | 195.31 | 69 | 234.16 | 39 |
32 | 1 | 5 | 0.05 | 179.33 | 362 | 204.72 | 72 | 251.48 | 30 |
32 | 1 | 5 | 0.1 | 184.27 | 384 | 217.90 | 50 | 223.01 | 50 |
32 | 1 | 5 | 0.15 | 205.24 | 327 | 228.40 | 80 | 223.52 | 81 |
32 | 1 | 10 | 0 | 264.73 | 196 | NA | NA | NA | NA |
32 | 1 | 10 | 0.05 | 231.74 | 302 | NA | NA | NA | NA |
32 | 1 | 10 | 0.1 | 196.73 | 399 | NA | NA | NA | NA |
32 | 1 | 10 | 0.15 | 197.42 | 348 | NA | NA | NA | NA |
32 | 2 | 5 | 0 | 271.67 | 377 | 226.19 | 72 | 200.75 | 51 |
32 | 2 | 5 | 0.05 | 259.13 | 136 | 213.24 | 79 | 240.31 | 89 |
32 | 2 | 5 | 0.1 | 272.13 | 89 | 235.56 | 57 | 225.07 | 64 |
32 | 2 | 5 | 0.15 | 238.8 | 62 | 233.29 | 71 | 224.44 | 100 |
32 | 2 | 10 | 0 | 170.41 | 351 | NA | NA | NA | NA |
32 | 2 | 10 | 0.05 | 185.37 | 395 | NA | NA | NA | NA |
32 | 2 | 10 | 0.1 | 189.83 | 260 | NA | NA | NA | NA |
32 | 2 | 10 | 0.15 | 200.49 | 328 | NA | NA | NA | NA |
64 | 1 | 5 | 0 | 230.31 | 153 | 221.11 | 79 | 241.49 | 88 |
64 | 1 | 5 | 0.05 | 189.14 | 322 | 205.10 | 72 | 214.19 | 40 |
64 | 1 | 5 | 0.1 | 255.40 | 307 | 222.80 | 66 | 253.20 | 51 |
64 | 1 | 5 | 0.15 | 245.71 | 53 | 207.48 | 78 | 251.50 | 63 |
64 | 1 | 10 | 0 | 193.33 | 391 | NA | NA | NA | NA |
64 | 1 | 10 | 0.05 | 198.98 | 142 | NA | NA | NA | NA |
64 | 1 | 10 | 0.1 | 210.31 | 351 | NA | NA | NA | NA |
64 | 1 | 10 | 0.15 | 191.81 | 399 | NA | NA | NA | NA |
64 | 2 | 5 | 0 | 314.57 | 140 | 207.66 | 67 | 219.21 | 100 |
64 | 2 | 5 | 0.05 | 314.30 | 141 | 227.20 | 61 | 212.24 | 99 |
64 | 2 | 5 | 0.1 | 278.63 | 386 | 240.29 | 71 | 218.14 | 67 |
64 | 2 | 5 | 0.15 | 293.13 | 126 | 236.68 | 78 | 227.72 | 77 |
64 | 2 | 10 | 0 | 220.30 | 356 | NA | NA | NA | NA |
64 | 2 | 10 | 0.05 | 192.39 | 340 | NA | NA | NA | NA |
64 | 2 | 10 | 0.1 | 218.01 | 365 | NA | NA | NA | NA |
64 | 2 | 10 | 0.15 | 216.56 | 349 | NA | NA | NA | NA |
Parameters | FNN Results | GRU Results | LSTM Results | |||||
---|---|---|---|---|---|---|---|---|
nnodes | nlayers | dropout | MAE | epoch | MAE | epoch | MAE | epoch |
32 | 1 | 0 | 323.95 | 83 | 269.42 | 71 | 265.85 | 57 |
32 | 1 | 0.1 | 387.77 | 164 | 251.01 | 41 | 276.99 | 34 |
32 | 1 | 0.2 | 409.22 | 174 | 281.80 | 76 | 267.53 | 55 |
32 | 2 | 0 | 243.30 | 352 | 251.17 | 53 | 305.92 | 97 |
32 | 2 | 0.1 | 266.86 | 304 | 278.23 | 67 | 274.52 | 97 |
32 | 2 | 0.2 | 276.40 | 349 | 280.47 | 99 | 265.23 | 52 |
32 | 3 | 0 | 227.40 | 96 | 284.14 | 97 | 306.19 | 58 |
32 | 3 | 0.1 | 232.65 | 374 | 293.56 | 98 | 281.66 | 38 |
32 | 3 | 0.2 | 273.96 | 209 | 274.05 | 73 | 275.03 | 99 |
64 | 1 | 0 | 339.85 | 59 | 263.20 | 82 | 284.72 | 68 |
64 | 1 | 0.1 | 327.44 | 51 | 275.22 | 99 | 319.13 | 19 |
64 | 1 | 0.2 | 388.25 | 68 | 290.82 | 97 | 269.69 | 43 |
64 | 2 | 0 | 224.12 | 324 | 243.72 | 56 | 234.22 | 99 |
64 | 2 | 0.1 | 277.33 | 289 | 281.96 | 77 | 254.63 | 97 |
64 | 2 | 0.2 | 311.62 | 369 | 296.06 | 80 | 279.30 | 92 |
64 | 3 | 0 | 237.82 | 180 | 266.50 | 95 | 296.79 | 88 |
64 | 3 | 0.1 | 279.02 | 288 | 286.08 | 92 | 281.90 | 87 |
64 | 3 | 0.2 | 296.57 | 66 | 290.56 | 93 | 260.95 | 100 |
Model | MAPE | Prediction horizon |
Data source | Published year | Reference |
ANN model | 2.90% | 1hr | DSO, Delhi, India | 2018 | Selvi et al. [16] |
1.96% | 1hr | Bandar Abbas, Iran | 2012 | Torabi et al. [22] | |
CNN-LSTM | 2.02% | 1hr | Public dataset, England,USA | 2019 | Pramono et.al. [23] |
34.84% | 1hr | UCI ML dataset (households) | 2019 | Kim et al. [24] | |
1% | 24hr | Industrial area, China | 2020 | Qi et al. [25] |
Method | Result | Reference |
MLR with AR(2) | Bayesian estimation provides consistent and better accuracy compared to OLS estimation | [32] |
PSO with ANN | Implementing PSO on ANN model outperformed shallow ANN model | [46] |
OLS | Interation of variable improves the prediction accuracy | [31] |
OLS and Bayesian estimation | Including temperature variable in a model can improved the prediction accuracy upto 20% | [45] |
PSO & GA with ANN | PSO+GA outperformed PSO with ANN | [35] |
OLS, GLSAR, FF-ANN | OLS and GLSAR models showed better forecasting accuracy than FF-ANN | [36] |
Ensemble for regression and ML | Lowers the test MAPE implementing blocked Cross Validation scheme. | [37] |
FNN, RNN based LSTM & GRU | For weekdays and for aggregate data GRU shows better accuracy | In this study |
Types | Variables | Description |
Deterministic | WD | Week dummy [Mon <Tue ... <Sat<Sun] |
MD | Month dummy [Feb <Mar <... <Nov <Dec] | |
DayAfterHoliday | Binary 0 or 1 | |
DayAfterLongHoliday | Binary 0 or 1 | |
DayAfterSongkran | Binary 0 or 1 | |
DayAfterNewyear | Binary 0 or 1 | |
Temperature | Temp | Forecasted temperature |
MaxTemp | Maximum forecasted temperature | |
Square temperature | Square of the forecasted temperature | |
MA2pmTemp | Moving avearage of temperature at 2pm | |
Lagged | load1d_cut2pm | 1 day ahead untill 2pm and 2 day ahead after 2pm load |
load2d_cut2pm | 2 days ahead untill 2pm and 3 day ahead after 2pm load | |
load3d_cut2pmR | 3 days ahead untill 2pm and 4 days ahead after 2pm load | |
load4d_cut2pmR | 4 days ahead untill 2pm and 5 days ahead after 2pm load | |
Interaction | WD:Temp | Interaction: week day dummy to temperature |
MD:Temp | Interaction: month dummy to temperature | |
WD:load1d_cut2pm | Interaction: week day dummy to load1d_cut2pm | |
WD:load2d_cut2pm | Interaction: week day dummy to load2d_cut2pm |
![]() |
Daytype | FNN | GRU | LSTM |
non-holiday weekdays | 2.97 | 2.71 | 3.76 |
non-holiday weekends | 3.83 | 4.62 | 3.58 |
holidays | 9.79 | 6.70 | 6.96 |
Overall | 3.54 | 3.44 | 3.86 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
Altmetrics
Downloads
157
Views
135
Comments
0
Subscription
Notify me about updates to this article or when a peer-reviewed version is published.
© 2025 MDPI (Basel, Switzerland) unless otherwise stated