Research Article
A Simple Method of Residential Electricity Load Forecasting by Improved Bayesian Neural Networks
Table 3
Results of IBNN models with different range of history data.
| | Model IBNN (Inputs number/Hidden neurons) | Training set (60%) | Test set (20%) | Computing time (s) | | MSE | R | MAPE (%) | MSE | R | MAPE (%) |
| | BNN_0.5hours (9/8) | 9.83e-2 | 8.71e-1 | 11.58 | 9.43e-2 | 8.71e-1 | 11.66 | 9 | | BNN_1hour (10/8) | 9.63e-2 | 8.72e-1 | 11.30 | 9.54e-2 | 8.76e-1 | 11.88 | 10 | | BNN_2hours (12/8) | 9.46e-2 | 8.77e-1 | 11.45 | 1.00e-1 | 8.58e-1 | 12.32 | 14 | | BNN_2days (13/8) | 8.83e-2 | 8.85e-1 | 10.80 | 8.62e-2 | 8.84e-1 | 10.36 | 14 | | BNN_3days (14/8) | 8.66e-2 | 8.86e-1 | 10.65 | 8.45e-2 | 8.91e-1 | 10.60 | 13 | | BNN_1week (15/8) | 8.37e-2 | 8.92e-1 | 10.64 | 8.97e-2 | 8.76e-1 | 10.68 | 16 | | BNN_2weeks (16/8) | 8.43e-2 | 8.89e-1 | 10.63 | 8.57e-2 | 8.89e-1 | 10.54 | 18 | | BNN_1month (18/8) | 8.30e-2 | 8.92e-1 | 10.81 | 7.96e-2 | 8.92e-1 | 10.35 | 34 | | BNN_2months (22/8) | 7.44e-2 | 9.05e-1 | 10.67 | 8.30e-2 | 8.85e-1 | 11.44 | 37 |
|
|
