I am having troubles with the following script for a non linear regression. Any help would be much appreciated. (The M values are very small)
X = [vi,vi,Mpost,Mpre] y = Q V=950*10^(-6) i=6*10^(-6) modelfun = @(b,x) V*b(1)-((((x(:,3))*((x(:,1))))-(1-(i/V))*(x(:,4))*(x(:,2)))) beta0 = [-6000] mdl = fitnlm(X,y,modelfun,beta0)
X =
4.5853 4.5853 0.0000 0
4.5681 4.5681 0.0000 0.0000
4.5509 4.5509 0.0000 0.0000
4.5336 4.5336 0.0000 0.0000
4.5163 4.5163 0.0000 0.0000
4.4989 4.4989 0.0000 0.0000
4.4815 4.4815 0.0000 0.0000
4.4641 4.4641 0.0000 0.0000
4.4467 4.4467 0.0000 0.0000
4.4293 4.4293 0.0000 0.0000
4.4118 4.4118 0.0000 0.0000
4.3943 4.3943 0.0000 0.0000
4.3767 4.3767 0.0000 0.0000
4.3592 4.3592 0.0000 0.0000
4.3416 4.3416 0.0000 0.0000
4.3240 4.3240 0.0000 0.0000
4.3064 4.3064 0.0000 0.0000
4.2887 4.2887 0.0000 0.0000
4.2711 4.2711 0.0000 0.0000
4.2534 4.2534 0.0000 0.0000
4.2357 4.2357 0.0000 0.0000
4.2180 4.2180 0.0000 0.0000
4.2002 4.2002 0.0000 0.0000
4.1825 4.1825 0.0000 0.0000
4.1647 4.1647 0.0000 0.0000
4.1469 4.1469 0.0000 0.0000
y =
-415.0974
-403.3474
-412.2701
-398.9688
-395.1909
-376.4527
-344.7533
-315.6087
-279.5409
-227.0259
-161.0164
-126.6837
-104.0480
-63.0274
-53.1502
-40.5565
-40.5068
-37.6305
-31.9914
-25.5591
-21.4596
-28.7419
-27.9112
-25.2559
-22.9237
-18.1073
L =
M =
V =
i =
modelfun =
@(b,x)V*b(1)-((((x(:,3))*((x(:,1))))-(1-(i/V))*(x(:,4))*(x(:,2))))
beta0 =
Error using nlinfit (line 205) Error evaluating model function '@(b,x)V*b(1)-((((x(:,3))*((x(:,1))))-(1-(i/V))*(x(:,4))*(x(:,2))))'.
Error in NonLinearModel/fitter (line 1122) [model.Coefs,~,J_r,model.CoefficientCovariance,model.MSE,model.ErrorModelInfo,~] = ...
Error in classreg.regr.FitObject/doFit (line 220) model = fitter(model);
Error in NonLinearModel.fit (line 1430) model = doFit(model);
Error in fitnlm (line 94) model = NonLinearModel.fit(X,varargin{:});
Caused by: Error using * Inner matrix dimensions must agree.
trying to do it as a table yields the same result: >> tlb=table(vi,vi1,Mpost,Mpre,Q) L=89*10^(-6) M=11*10^(-6) V=950*10^(-6) i=6*10^(-6) modelfun = @(b,x)V*b(1)*(x(:,3)*x(:,1)-(1-(i/V))*x(:,4)*x(:,2)) beta0 = [-6000] mdl = fitnlm(tlb,modelfun,beta0)
tlb =
vi vi1 Mpost Mpre Q
______ ______ __________ __________ _______
4.5853 4.6026 6.9474e-08 0 -415.1
4.5681 4.5853 1.3851e-07 6.9474e-08 -403.35
4.5509 4.5681 2.0711e-07 1.3851e-07 -412.27
4.5336 4.5509 2.7527e-07 2.0711e-07 -398.97
4.5163 4.5336 3.4301e-07 2.7527e-07 -395.19
4.4989 4.5163 4.1032e-07 3.4301e-07 -376.45
4.4815 4.4989 4.772e-07 4.1032e-07 -344.75
4.4641 4.4815 5.4366e-07 4.772e-07 -315.61
4.4467 4.4641 6.097e-07 5.4366e-07 -279.54
4.4293 4.4467 6.7532e-07 6.097e-07 -227.03
4.4118 4.4293 7.4053e-07 6.7532e-07 -161.02
4.3943 4.4118 8.0533e-07 7.4053e-07 -126.68
4.3767 4.3943 8.6971e-07 8.0533e-07 -104.05
4.3592 4.3767 9.3369e-07 8.6971e-07 -63.027
4.3416 4.3592 9.9727e-07 9.3369e-07 -53.15
4.324 4.3416 1.0604e-06 9.9727e-07 -40.557
4.3064 4.324 1.1232e-06 1.0604e-06 -40.507
4.2887 4.3064 1.1856e-06 1.1232e-06 -37.631
4.2711 4.2887 1.2476e-06 1.1856e-06 -31.991
4.2534 4.2711 1.3092e-06 1.2476e-06 -25.559
4.2357 4.2534 1.3704e-06 1.3092e-06 -21.46
4.218 4.2357 1.4312e-06 1.3704e-06 -28.742
4.2002 4.218 1.4916e-06 1.4312e-06 -27.911
4.1825 4.2002 1.5517e-06 1.4916e-06 -25.256
4.1647 4.1825 1.6114e-06 1.5517e-06 -22.924
4.1469 4.1647 1.6707e-06 1.6114e-06 -18.107
L =
M =
V =
i =
modelfun =
@(b,x)V*b(1)*(x(:,3)*x(:,1)-(1-(i/V))*x(:,4)*x(:,2))
beta0 =
Error using nlinfit (line 205) Error evaluating model function '@(b,x)V*b(1)*(x(:,3)*x(:,1)-(1-(i/V))*x(:,4)*x(:,2))'.
Error in NonLinearModel/fitter (line 1122) [model.Coefs,~,J_r,model.CoefficientCovariance,model.MSE,model.ErrorModelInfo,~] = ...
Error in classreg.regr.FitObject/doFit (line 220) model = fitter(model);
Error in NonLinearModel.fit (line 1430) model = doFit(model);
Error in fitnlm (line 94) model = NonLinearModel.fit(X,varargin{:});
Caused by: Error using * Inner matrix dimensions must agree.
Best regards: Emil Stender
