Plot deviation of curve as shaded area inside

Hi,

A common question here in this forum is how to plot an area between two curves.

Here's one type of a common answer:

x = transpose(1:100);
y = x.^2;

curve1 = y*1.05;
curve2 = y*0.95;
x2 = [x flipud(x)];
inBetween = [curve1, curve2];
fill(x2, inBetween, 'k', 'LineStyle', 'none');
hold on;
plot(x, y, 'r', 'LineWidth', 2);

This is the result I get:

I used

alpha(.1)

to make the shaded area transparent.

What I would like to have is the area between those two curves shaded, since I want to show the red curve's deviation without having to use the green lines. Hence why I removed the Linestyle. But I get something like a linear fill between the starting and endpoint of the vectors.

How can I do that without too many lines of code? I'm afraid my laptop is at a certain limit with the actual length of vectors I have to plot (>800.000 values).

Also if you know if and how it's possible to alternate the shaded area like a zebra (white-black-white-black, etc) I'd appreciate if you tell me that!

Thank you!

Answer 1

[Star Strider]

I generally prefer patch to fill —

x = transpose(1:100);
y = x.^2;

curve1 = y*1.05;
curve2 = y*0.95;
x2 = [x flipud(x)];
inBetween = [curve1  curve2];
patch([x; flip(x)], [inBetween(:,1); flip(inBetween(:,2))], 'k', 'LineStyle', 'none', 'FaceAlpha',0.25);
hold on;
plot(x, y, 'r', 'LineWidth', 2);

The approach to writing patch calls is to provide a defined area that the function can fill. Here, that is provided by plotting ‘curve1’ against ‘x’ and then plotting ‘flip(curve2)’ against ‘flip(x)’ completing the region to be filled.

.

[Sokratis Panagiotidis]

Hey man thanks a lot!

This is what I was looking for.

[Star Strider]

As always, my pleasure!

Answer 2

[Bjorn Gustavsson]

Have a look at the file exchange. There there are multiple "shaded-error-plots". For example (without ordering of merit):

jackknife, errorshade, errorfill, shadederrorbar and confplot.

I have used a majority of these functions at different times to god result, now I have no memory of why one or the other was preferable.

HTH

[Sokratis Panagiotidis]

The 'errorfill' was pretty good and what I was actually looking for, but it wouldn't work with the vectors I wanted to plot (the x.^2 was just an example, my actual vector is taken from an experiment I did).

Still tanks a lot!

[Bjorn Gustavsson]

In what ways were those functions unsuitable for your data?

[Sokratis Panagiotidis]

This is what I get when using errorfill:

Unable to perform assignment because the indices on the left side are not compatible with the
size of the right side.

Error in errorfill (line 111)
E(:,n)=		[y+temp(1,:)     y(end:-1:1)-temp(2,end:-1:1)];

Error in Errorfill_Test (line 7)
errorfill(x, y, E3, '.');

My vector has almost 600 thousand entries so it's hard to tell what the problem is.