how to remove the distortion by the end of the graph ?

[Walter Roberson]

We here have no information as to what part of it is distortion and what part of it is valid data.

[shrik kader]

Walter Roberson, the end of the graph from the right is not needed, the curve is only needed here. P_V curve of PV characteristics.

[Walter Roberson]

Tell us a rule that allows us to determine whether a particular portion of the graph is "needed" or not. For example, "10" is "from the right" compared to "5", so should we just tell you how to end your plot at 5? Or 2? Or should it end at roughly 27 at the location where it starts turning down?

[shrik kader]

after the 34, the curve should turn down smoothly as a usual curve without the distortion after 35 and ends at 36.3.

[John D'Errico]

So just delete all points where x is greater than 34. WTP?

[shrik kader]

the graph will not take its normal shape, how to delete them ?

Answer 1

[John D'Errico]

Sigh. What do you mean it will not take its normal shape?

y(x > 34) = [];
x(x > 34) = [];

[shrik kader]

Did not work, gave me the same graph.

[John D'Errico]

Well, since it DOES work, you need to show what you did. Did you honestly expect that merely deleting those values would automatically make the graph update itself?

After deleting the garbage, replot the vectors.

plot(x,y)

[shrik kader]

this is the picture after deleting the garbage,some portion of the curve is missing. how can I make it end at the axis with no distortion.

[Image Analyst]

Why don't we just speed up the solution by you attaching your data?

Or try getting rid of data more than 100

y(y > 100) = [];
x(y > 100) = [];
plot(x, y, 'r-');

but seriously, attaching your data would be fastest. Read this

[Walter Roberson]

You appear to be asking that somehow an undistorted line is plotted.

Have you considered using a low-pass filter?

Have you considered determining a mathematical model for what your graph should look like, and then using the known points to determine the parameters of the model, and then using the model to project data?

Have you considered taking the points between the peak and the beginning of the distorted area, and fitting a low degree polynomial to it, and using that polynomial to project to the end of the area?