Research Article

Nonlinear Model Order Selection: A GMM Clustering Approach Based on a Genetic Version of EM Algorithm

Table 9

RMSE prediction errors measured by five methods of case 3.

Ex. 3 RMSE
Input order (n)
Output order (m)
01234

RBF00.09130.08050.12930.2905
10.09010.09000.08190.18340.3411
20.07840.08030.08020.27330.7444
30.07400.08150.12090.25760.9637
40.26860.36210.47030.57110.8461
SVR01.26970.11790.16100.3664
11.78951.92770.15250.10620.3198
20.57620.11380.10550.11430.3730
30.12750.15600.13030.14990.3435
40.32510.33000.33200.29960.3564

BA-SVR00.18260.11630.11980.1441
10.42110.14350.10450.09910.1021
20.26750.10210.09590.10650.1152
30.12200.13250.11550.13900.1707
40.31670.11560.13090.18550.1893

LM-BP01.23960.08890.08980.1068
10.10140.11110.09200.08970.1247
20.13670.09040.08670.09090.0901
30.11760.10660.09240.10080.1179
40.08700.10140.09020.10150.0935

ELM00.04140.02560.02440.0271
10.01280.01560.02160.02540.0231
20.01230.02270.01960.03800.0291
30.01760.02300.01970.02410.0223
40.03530.02790.02470.01770.0277