using nonlinearmodek fit for the first time

Hi,

I wanted to fit a nonlinear model to my data but I keep getting an error.

This is the error:

>> untitled2

Unable to resolve the name NonLinearModel.fit.

Error in untitled2 (line 17)

myFit = NonLinearModel.fit(X,Y, 'y ~ b0 + b1*cos(b2*x1 + b3)*exp(-b4*x1)', [B0, B1, B2, B3,B4]);

This is my code:

X=table2array(X);

Y=table2array(Y);

% Note that we need to pass three sets of input arguments to NonLinearModel

% # The X and Y data

% # A string describing our model

% # Starting conditions for the optimization solvers

% Generate some good starting conditions for the solvers

scatter(X, Y)

hold on

B0 = mean(Y); % Vertical shift

B1 = (max(Y) - min(Y))/2; % Amplitude

B2 = 2; % Phase (Number of peaks)

B3 = 0; % Phase shift (eyeball the Curve)

B4 = 5;

myFit = NonLinearModel.fit(X,Y, 'y ~ b0 + b1*cos(b2*x1 + b3)*exp(-b4*x1)', [B0, B1, B2, B3,B4]);

% Note that all the coefficient estimates are very good except for b3 where

% any even integer is equally valid

%%look at the complete set of methods

methods(myFit)

%%Generate a plot

hold on

plot(X, myFit.Fitted)

hold off

Do I need to define the nonlinearmodel? or download another extent?

Thank you so much in advanced!

Answer 1

[Star Strider]

The NonLinearModel is a class.

To do the actual nonlinear regression, use the fitnlm function.

[Michal Naim Ben Eliyahu]

is there an exemple tohow to use it?

I've never used it before and the info page is unclear to me.

do I create the model class by using the function?

thank you so much!

[Star Strider]

Try something like this —

X = (0:15).';
Y = exp(-0.2*X) .* sin(2*pi*X/5) + randn(size(X))/5;

figure
plot(X,Y)
grid

B0 = mean(Y);  % Vertical shift
B1 = (max(Y) - min(Y))/2; % Amplitude
B2 = 2; % Phase (Number of peaks)
B3 = 0; % Phase shift (eyeball the Curve)
B4 = 0.1;

myFit = fitnlm(X,Y, 'y ~ b0 + b1*cos(b2*x1 + b3)*exp(-b4*x1)', [B0, B1, B2, B3,B4])

myFit = 
Nonlinear regression model:
    y ~ b0 + b1*cos(b2*x1 + b3)*exp( - b4*x1)

Estimated Coefficients:
          Estimate        SE        tStat        pValue  
          _________    ________    ________    __________

    b0    0.0010881    0.040724    0.026719       0.97916
    b1     -0.92925     0.18102     -5.1335    0.00032654
    b2       1.2567     0.03352      37.492    5.8692e-13
    b3       1.3098     0.14409      9.0897    1.9022e-06
    b4      0.13028      0.0405      3.2169     0.0082047


Number of observations: 16, Error degrees of freedom: 11
Root Mean Squared Error: 0.157
R-Squared: 0.854,  Adjusted R-Squared 0.801
F-statistic vs. constant model: 16.1, p-value = 0.000145

[yfit,yfitci] = predict(myFit,X)

yfit = 16×1

-0.2387 0.6856 0.5572 -0.2248 -0.5500 -0.1238 0.3580 0.2909 -0.1167 -0.2862

yfitci = 16×2

-0.5321 0.0547 0.4021 0.9691 0.3432 0.7712 -0.4193 -0.0302 -0.7311 -0.3688 -0.2879 0.0404 0.1837 0.5323 0.1146 0.4672 -0.2873 0.0538 -0.4579 -0.1144

figure
plot(X, Y, 'xb')
hold on
plot(X, yfit, '-r')
plot(X, yfitci, '--r')
hold off
grid
legend('Data','Regression','±95% CI', 'Location','best')

I changed some of the initial parameter estimates to work with my simulated data.

.

[Michal Naim Ben Eliyahu]

sorry for keep asking but this time I used fitnlm and I keep getting a straight line as my fit and changing my parameters doesn't affect anything.

this is the code this time:

X=data(100:250,1);

Y=data(100:250,2);

X=table2array(X);

Y=table2array(Y);

% Note that we need to pass three sets of input arguments to NonLinearModel

% # The X and Y data

% # A string describing our model

% # Starting conditions for the optimization solvers

% Generate some good starting conditions for the solvers

scatter(X, Y)

hold on

B0 = mean(Y); % Vertical shift

B1 = (max(Y) - min(Y))/2; % Amplitude

B2 = 2 ; % Phase (Number of peaks)

B3 = 0; % Phase shift (eyeball the Curve)

B4 = 5;

myFit = fitnlm(X,Y, 'y ~ b0 + b1*cos(b2*x1 + b3)*exp(-b4*x1)', [B0, B1, B2, B3,B4]);

yfit=predict(myFit,X);

% Note that all the coefficient estimates are very good except for b3 where

% any even integer is equally valid

%%look at the complete set of methods

%%methods(myFit)

%%Generate a plot

plot(X,Y,'xb')

hold on

plot(X,yfit)

hold off

and this is thr result:

[Star Strider]

Change ‘B4’ to something smaller (perhaps 2). It may also be necessary to subtract ‘X(1)’ from the rest of ‘X’ so that the record begins at 0. That should not affect the parameter estimates.

It may be necessary to experiment a bit, especially with ‘B4’.

It would help to have your data, preferably as a spreadsheet or text file since the online Run feature does not work well with .mat files.

That aside, my code works!

Running your data with the code from How to filter noise from time-frequency data and find natural frequency of a cantilever? could help with estimating the parameters. Then use those as the initial parameter estimates for fitnlm. It will improve on the initial estimates, and give all the necessary statistics.