how can i fit with using nonlinear regression?

hello everyone, i have data which is uploded on matlab but i dont have function about this data.I just plot the graph but i dont know how to code to get nonlinear regression conclusion on my graph.I share my question and graph.Please help me thanks a lot.

Question:Find the best fit line in the least squares sense to the training data. Plot the data and the best fit line on the same plot. Compute the correlation coefficient.How much of the original uncertainty is explained by the linear model?

Answer 1

[Star Strider]

There are two options —

First, if it is truly a nonlinear funciton of something, developa mathematical model of the process that created those data and then use it to estimate the parameters. I can help with that if necessary.

Second, if it is only necessary to fit the funciton to something, the easiest approach is to use polyfit and polyval.

.

[alp onur karacay]

thanks for quick answer,

Can you do it with first method? I share my matlab documents

thanks again.

[Star Strider]

What process created the data?

Is it possible to describe it, ideally mathematically? (The ‘lab3.m’ file appears to be an attempt at a second-order polynomial regression without an intercept term.)

Try this —

LD = load('proj3data (1).mat');
x = LD.x;
y = LD.y;

DM = [x(:).^2 x(:) ones(size(x(:)))];                       % Design Matrix
B = DM \ y(:)
yest = DM * B;

DMmdl = [x(:) x(:).^2];                                     % Design Matrix For 'fitlm'
lab3mdl = fitlm(DMmdl,y(:))

figure
plot(x, y, '.')
hold on
plot(x, yest, '-r', 'LineWidth',2)
hold off
grid
legend('Data','Regression Fit', 'Location','best')
text(mean(xlim), mean(ylim), sprintf('$$y(x) = %.3fx^2 + %.3fx + %.3f$$',B), 'HorizontalAlignment','center', 'Vert','bottom', 'Interpreter','latex')
figure
plot(lab3mdl)
grid

.

[alp onur karacay]

thank you so much it works

[Star Strider]

As always, my pleasure!

.