Multilevel Information Granule Construction Model Based on Large Data Environment and Its Application in Time Series

Table 1

Predicted results of Scheme 3.

Equivalent partition set	Machine method	Mining the maximum value of evaluation effect function (%)	Function maximum corresponds to time granule
L1 = {1, 2, 4}	SVM	4.39	180 hours
	ID3	5.13
	PNN	2.23
L2 = {1, 2, 4, 8}	SVM	28.84	90 hours
	ID3	28.91
	PNN	21.38
L3 = {1, 2, 4, 8, 16, 32}	SVM	49.21	45 hours
	ID3	49.32
	PNN	43.82
L4 = {1, 2, 4, 8, 16, 32, 64, 128, 256}	SVM	86.61	5.625 hours
	ID3	92.13
	PNN	83.43