Evaluation of Coalbed Methane Content by Using Kernel Extreme Learning Machine and Geophysical Logging Data

Table 2

Kernel function optimization results and error statistics of KELM method under various data ratios.

Ratio (training : testing)	Kernel function	Average relative error (%)
		Training dataset	Testing dataset	Error variation (%)
5 : 5	6.1	13.72	18.71	4.99
6 : 4	9.1	14.67	17.95	3.28
7 : 3	3.9	14.49	15.99	1.5
8 : 2	5.6	14.54	17.81	3.27
9 : 1	3.2	12.06	18.63	6.57

^aError variation = mean relative error difference between testing dataset and training dataset.