Smooth data to get the best approximation.

Hello everyone, I was wondering if anyone knows how I could process the input data in such a way as to obtain something similar to what I show in smooth data. The points marked in red and blue correspond to the change points when using the ischange tool

cI = ischange(YY,'MaxNumChanges',1);
scatter(XX(cI),YY(cI),'filled')

I include an attached data with the points shown in the graph being XX and YY the input data (the ones I intend to modify) and the XSmooth, YSmooth data the ones I intend to get.

I've tried with the methods include in the smoothdata function but i couldn't get the result i expected

methods = {'movmean','movmedian','gaussian','lowess','loess','rlowess','rloess','sgolay'};

Answer 1

[hosein Javan]

I used curve fitting app you can workaround and change settings to get your result.

[Alejandro Fernández]

The result remains unapproached regardless of the parameter you use to calculate the spline.

[Alejandro Fernández]

But thanks for your comment :D

[hosein Javan]

there is a sharp spike at x=120. you can apply curve fitting to each interval seperately that does not affect the whole interval. This surely is it.

[hosein Javan]

another method I'm thinking that your signal is full of noise. instead of curve fitting you can refer to signal processing toolbox see if there is a tool for cleaning up and denoising signals. consult with a communication expert. they can help you with frequency analysis. hope that helped.

[Alejandro Fernández]

Ok, thank you so much for all your help :D

[hosein Javan]

Alajendro Fernández De nada.

Answer 2

[Bruno Luong]

Here is what my ftting tool gives

load('InputData_SmoothData.mat')

pp=BSFK(XX,YY); % FEX file

plot(XX,YY);
plot(XX,ppval(pp,XX),'r');
legend('raw data','BSFK fitting')