Forecasting Electrical Energy Consumption of Equipment Maintenance Using Neural Network and Particle Swarm Optimization

Table 3

The MAPE (%) associated with the four approaches on the testing dataset of 2009.

Organization	ANN	ANN-PSO	R-PSO	ANN-APSO
#1	1.816^+	1.220^+	1.093^+	0.835
#2	1.493^+	0.000	0.355^+	0.000
#3	5.267^+	1.801^+	1.103	1.004
#4	2.205^+	0.934^+	0.783^+	0.297
#5	1.315^+	0.303^+	0.526^+	0.067
#6	3.655^+	1.500^+	1.218^+	0.722
#7	0.452^+	0.000	0.000	0.000

In column 2–4, ⁺means that the result of ANN-APSO has statistically significant improvement over that of the comparative algorithm (using paired one-tailed 𝑡-test at a 0.05 level of significance).