Very Short-Term Load Forecasting Using Hybrid Algebraic Prediction and Support Vector Regression

Table 3

Percentage error when each from the training set is used as an input to the trained SVR module.

Date			MW/C°	% error
	July 1	1341.633			1338.462	669.231	0.236
July 3	−4110.892	−4107.722	586.817	0.077
July 9	1642.079	1638.908	409.727	0.193
July 16	−818.390	−821.561	821.561	−0.387
July 17	−1285.362	−1282.191	641.096	0.247
July 23	−2228.795	−2225.624	1112.812	0.142
July 24	−915.093	−918.264	918.263	−0.346
July 25	3170.716	2116.684	351.949	0.1
July 28	−4190.624	−4187.454	1046.863	0.076
Aug 4	−7426.996	−4224.748	1237.304	0.043
Aug 12	2501.683	2116.684	1249.256	0.127
Aug 14	−3083.127	−3079.957	1026.652	0.103
Aug 27	−1438.378	−1435.207	717.603	0.220
Sept 1	2719.843	2116.684	1358.336	0.1175
Sept 2	−1321.296	−1318.126	659.063	0.240
Sept 12	−1582.431	−1579.260	1579.260	0.200
Sept 17	−2069.028	−2065.858	516.464	0.153