The SAS System 14:11 Thursday, March 23, 2006 1 The NEURAL Procedure Optimization Start Parameter Estimates Gradient Objective N Parameter Estimate Function 1 AGE_H11 -0.021490 0 2 BMI_H11 -0.049994 0 3 BP_H11 -0.121747 0 4 NPREG_H11 -0.111131 0 5 BIAS_H11 0.073533 0 6 H11_GLU 0 3.478848 7 BIAS_GLU 123.030000 2.273737E-15 Value of Objective Function = 961.4091 The SAS System 14:11 Thursday, March 23, 2006 2 The NEURAL Procedure Levenberg-Marquardt Optimization Minimum Iterations 0 Maximum Iterations 100 Maximum Function Calls 2147483647 Maximum CPU Time 600 ABSGCONV Gradient Criterion 0.00001 GCONV Gradient Criterion 1E-8 GCONV2 Gradient Criterion 0 ABSFCONV Function Criterion 0 FCONV Function Criterion 0.0001 FCONV2 Function Criterion 0 FSIZE Parameter 0 ABSXCONV Parameter Change Criterion 0 XCONV Parameter Change Criterion 0 XSIZE Parameter 0 ABSCONV Function Criterion 0.971120303 Trust Region Initial Radius Factor 1 Singularity Tolerance (SINGULAR) 1E-8 Levenberg-Marquardt Optimization Scaling Update of More (1978) Parameter Estimates 7 Optimization Start Active Constraints 0 Objective Function 961.4091 Max Abs Gradient Element 3.4788481671 Radius 1.0654668915 Ratio Between Actual Objective Max Abs and Function Active Objective Function Gradient Predicted Iter Restarts Calls Constraints Function Change Element Lambda Change 1* 0 2 0 949.86848 11.5406 71.9932 9.676 1.000 2 0 3 0 904.66864 45.1998 13.3657 11.946 0.361 3 0 4 0 880.68285 23.9858 11.6853 4.210 1.025 4 0 5 0 845.60538 35.0775 7.1606 1.202 1.009 5 0 7 0 826.78746 18.8179 3.9983 0.674 1.022 6 0 9 0 819.15690 7.6306 3.6259 0.640 0.996 7 0 10 0 814.45157 4.7053 26.4767 0.172 0.577 8 0 11 0 808.54224 5.9093 8.4994 0.0558 0.980 9 0 13 0 807.58228 0.9600 4.7206 0.0887 0.926 10 0 14 0 806.85410 0.7282 11.9295 0.0239 0.746 The SAS System 14:11 Thursday, March 23, 2006 3 The NEURAL Procedure Ratio Between Actual Objective Max Abs and Function Active Objective Function Gradient Predicted Iter Restarts Calls Constraints Function Change Element Lambda Change 11 0 15 0 806.41678 0.4373 11.5066 0.00765 0.760 12 0 16 0 806.30513 0.1117 14.5447 0 0.521 13 0 17 0 806.20642 0.0987 1.2139 0 0.969 14 0 19 0 806.20537 0.00105 0.7905 0.00150 0.406 15 0 21 0 806.20522 0.000146 0.3610 0.00070 0.171 16 0 22 0 806.20515 0.000073 0.3566 0.00164 0.105 17 0 24 0 806.20508 0.000068 0.3155 0.00321 0.133 18 0 25 0 806.20505 0.000029 0.3148 0.00292 0.0745 19 0 26 0 806.20499 0.000059 0.2521 0.00702 0.187 20 0 27 0 806.20497 0.000020 0.2298 0.00355 0.0953 21 0 28 0 806.20493 0.000048 0.1528 0.0181 0.311 22 0 29 0 806.20492 7.611E-6 0.1293 0.00366 0.111 23 0 30 0 806.20491 8.963E-6 0.1100 0.00596 0.164 Optimization Results Iterations 23 Function Calls 32 Jacobian Calls 25 Active Constraints 0 Objective Function 806.20490973 Max Abs Gradient Element 0.1100380562 Lambda 0.0059583611 Actual Over Pred Change 0.1638413628 Radius 0.0393148513 Convergence criterion (FCONV=0.0001) satisfied. NOTE: At least one element of the gradient is greater than 1e-3. The SAS System 14:11 Thursday, March 23, 2006 4 The NEURAL Procedure Optimization Results Parameter Estimates Gradient Objective N Parameter Estimate Function 1 AGE_H11 -0.394423 -0.110038 2 BMI_H11 -0.164447 0.070036 3 BP_H11 0.154887 -0.003872 4 NPREG_H11 -0.293413 0.038998 5 BIAS_H11 -0.566524 0.002489 6 H11_GLU -33.195684 -0.000195 7 BIAS_GLU 109.624549 -0.000122 Value of Objective Function = 806.20490973 ************************************************************************** **************************************************************************