Table 7 The results of equivalence tests for the best predictive DLA of the number of neuron alternatives regarding the numbers of hidden layer and M5 function based on NLRM, M5 based on NLME with f random, FFB-ANN, CC-ANN
The number of neuron alternatives and regression models | The number of hidden layers | b0 limits | Bootstrap b0 limits | b1 limits | Bootstrap b1 limits | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
lower | upper | lower | upper | H0: not Equivalent | lower | upper | lower | upper | H0: not Equivalent | ||
50 # neuron | 3 | 9.7579 | 11.9263 | 10.7578 | 10.9326 | Rejected | 0.9000 | 1.1000 | 0.90224 | 0.9871 | Rejected |
90 # neuron | 4 | 9.7579 | 11.9263 | 10.7458 | 10.9343 | Rejected | 0.9000 | 1.1000 | 0.9112 | 0.9895 | Rejected |
90 # neuron | 5 | 9.7579 | 11.9263 | 10.7492 | 10.9279 | Rejected | 0.9000 | 1.1000 | 0.9227 | 0.9886 | Rejected |
100 # neuron | 6 | 9.7579 | 11.9263 | 10.7494 | 10.9343 | Rejected | 0.9000 | 1.1000 | 0.9136 | 0.9892 | Rejected |
70 # neuron | 7 | 9.7579 | 11.9263 | 10.7530 | 10.9229 | Rejected | 0.9000 | 1.1000 | 0.9228 | 0.9941 | Rejected |
90 # neuron | 8 | 9.7579 | 11.9263 | 10.7431 | 10.9346 | Rejected | 0.9000 | 1.1000 | 0.8774 | 0.9829 | Not rejected |
100 # neuron | 9 | 9.7579 | 11.9263 | 10.7558 | 10.9357 | Rejected | 0.9000 | 1.1000 | 0.9015 | 0.9761 | Rejected |
80 # neuron | 10 | 9.7579 | 11.9263 | 10.7523 | 10.9339 | Rejected | 0.9000 | 1.1000 | 0.9178 | 0.9999 | Rejected |
M5 function based on NLRM | 9.7579 | 11.9263 | 10.7621 | 10.9238 | Rejected | 0.9000 | 1.1000 | 0.9851 | 1.0562 | Rejected | |
NLME with f random | 9.7579 | 11.9263 | 10.7655 | 10.9135 | Rejected | 0.9000 | 1.1000 | 0.9849 | 1.0591 | Rejected | |
FFB-ANN | 9.7579 | 11.9263 | 10.7400 | 10.9397 | Rejected | 0.9000 | 1.1000 | 0.9049 | 1.0184 | Rejected | |
CC-ANN | 9.7579 | 11.9263 | 10.7490 | 10.9354 | Rejected | 0.9000 | 1.1000 | 0.8686 | 1.0118 | Not rejected | |