Hybrid Machine Learning Model for Electricity Consumption Prediction Using Random Forest and Artificial Neural Networks

Table 4

Feature importance based on the SWR model.

Group of variables	Variable code		Coefficients	SE coef	t-value	-value	VIF
Geospatial	Usage_Type (dummy variable)	Agriculture	−3.724E + 04	1.144E + 03	-32.556	<2E − 16	1.7382
		Government	−3.755E + 04	1.146E + 03	-32.763	<2E − 16	1.7332
		House < 150	4.953E + 04	1.169E + 03	42.378	<2E − 16	1.6988
		House > 150	2.475E + 05	1.164E + 03	212.582	<2E − 16	1.7033
		Large	5.338E + 05	1.170E + 03	456.169	<2E − 16	1.6971
		Medium	1.781E + 05	1.170E + 03	152.206	<2E − 16	1.6966
		Small	1.029E + 05	1.149E + 03	89.538	<2E − 16	1.7275
	TOU (dummy variable)	Semipeak	1.687E + 04	7.327E + 02	23.031	<2E − 16	1.4001
		Peakday	2.737E + 04	1.044E + 03	26.210	<2E − 16	1.4000
Geographical	Population_Ratio		−1.330E + 02	1.417E + 00	-93.872	<2E − 16	3.2135
Climatic	Mean_MSL_Pressure		1.585E + 04	2.453E + 02	64.603	<2E − 16	4.4071
	Mean_Maximum_ temperature		2.438E + 04	3.329E + 02	73.236	<2E − 16	3.7922
	Mean_Relative_ humidity		7.057E + 03	1.073E + 02	65.774	<2E − 16	5.4626
	Total_Rainfall		1.114E + 01	4.604E-01	24.193	<2E − 16	2.9923
Industrial	Industrial_Labor		1.626E + 01	1.460E-01	111.316	<2E − 16	9.8164
	Industrial_Plant		−3.204E + 01	5.090E + 00	-6.294	3.09 E − 10	8.5413
Household	Household		1.059E + 02	1.144E + 00	92.641	<2E − 16	1.5609
	Liabilities		3.041E − 02	1.403E − 03	21.665	<2E − 16	2.3207