高手帮忙解释下bp神经网络的结果

那个高手能解释下用sas做bp神经网络的结果解释啊？以及有些指标的值取值为多少时，比较好

Label	Train	Valid	Test
[ TARGET=TYPE ]
Average Profit	0.5	0.485714286	0.457142857
Misclassification Rate	0.495238095	0.514285714	0.542857143
Average Error	0.134431584	0.137557174	0.137147385
Average Squared Error	0.134431584	0.137557174	0.137147385
Sum of Squared Errors	141.1531635	48.14501105	48.00158467
Root Average Squared Error	0.36664913	0.37088701	0.370334153
Root Final Prediction Error	0.393844343
Root Mean Squared Error	0.380489783	0.37088701	0.370334153
Error Function	141.1531635	48.14501105	48.00158467
Mean Squared Error	0.144772475	0.137557174	0.137147385
Maximum Absolute Error	0.823276371	0.826041771	0.829326746
Final Prediction Error	0.155113366
Divisor for ASE	1050	350	350
Model Degrees of Freedom	60
Degrees of Freedom for Error	780
Total Degrees of Freedom	840
Sum of Frequencies	210	70	70
Sum Case Weights * Frequencies	1050	350	350
Akaike's Information Criterion	-1378.187313
Schwarz's Baysian Criterion	-1094.1832

既然没人回，结贴吧

晕 这个不能算是神经网络的计算结果 你是直接用EM模块做的 在这里Average Error比较有用 可以根据Average Error值调整参数