Risk Prediction by Using Artificial Neural Network in Global Software Development

Table 4

Comparison of LM, SCG, and BR of time-related risks.

	Levenberg–Marquardt	Scaled Conjugate Gradient	Bayesian Regularization
Epoch	5 iteration	19 iteration	150 iteration
Processing	0 : 00 : 00	0 : 00 : 00	0 : 00 : 02
Performance	8.27	8.71	8.1
Gradient	0	3.16	0.0555
MU	0		5.00 E+10
MSE (validation)	8.02	9.08	2.157
R (validation)	0.85	0.84	0.9245
MSE (testing)	6.90	11.25	7.47
R (testing)	0.88	0.88	0.89