How to "force" a data to start and end at zero y values?

Hi!

How can I can force the end point to zero?

Red: original data (after cumtrapz);

Blue: filtered data (hp);

But I know that the data has to end at zero y value. Is there any way that I can "rotate" the blue line to fit? I looked at fit and polyfit, but I might doing something wrong.

Answer 1

[Image Analyst]

Just use polyfit on the first and last points to get a ramp. Then subtract the ramp from all the y values. The first and last point will subtract to give 0 and all the other points in between will be proportionally lowered.

coefficients = polyfit([x(1), x(end)], [y(1), y(end)], 1)
ramp = polyval(coefficients, x);
yFit = y - ramp;

[Vitor Sotero]

Wowww!

Thanks a lot! I spent so much time looking for this answer.

Thanks again!

[Image Analyst]

You're welcome. Could you then "Accept this answer"? Thanks in advance.

[Vitor Sotero]

Done!

[Vitor Sotero]

Image Analyst, sorry for one more question...

Can I do the same approach in this case?

Note that now I don't want to "force the data to zero. In this example, I would like that the point 4148 goes near the point 4149.

Thanks in advance!

[Image Analyst]

How do you want that to happen? Do you just want to clip it, like

index = find(x <= 4148, 1, 'last')
clipValue = y(index+1); % or mean(y(index+1:end) or min(y(index+1:end) or whatever you want.
v = 1 : length(y);
indexesToChange = y < clipValue & v > 4148;
y(indexesToChange) = clipValue;

[Vitor Sotero]

I want something like this:

[Image Analyst]

Simply use rescale() to rescale your data in the indexes from the peak to the min to go from the peak to the clipValue.

index = find(x <= 4148, 1, 'last')
[maxValue, indexAtMaxValue] = max(y);
minValue = y(index+1); % or mean(y(index+1:end) or min(y(index+1:end) or whatever you want.
y(indexAtMaxValue : index) = rescale(y(indexAtMaxValue : index), minValue, maxValue);

[Vitor Sotero]

Didn't work for me...

In this example, the blue line is my reference data, the black line the data that I want to fix and the red line is the result after using rescale...

Giving a zoom, I will show what I expected (the green line made by hand):

[Image Analyst]

I'm sorry but I can't find your data anywhere. Did you detach it? Where is it?

Please attach it (again) so I can try things.

[Vitor Sotero]

Here! data.mat contains my data on the first column and the reference data on the second. I attached segments.mat too, that contains the index.

In this picture, the black line is my data and the blue one is the reference.

"Why not just subtract the difference?" If I do this, the result is the red line:

[Vitor Sotero]

Any light?