# How can I plot distribution with shades?

1 view (last 30 days)
Giorgi on 18 Sep 2015
Commented: Kelly Kearney on 18 Sep 2015
Hello Matlabers :)
I have timeseries that I forecast for 12 period for 1000 times so I have 12X1000 matrix now and I want to make chart like this Can anyone help me?

Joseph Cheng on 18 Sep 2015
you'd do something like this where you would use fill to generate the ranges of the distribution
t = 0:.1:10;
x = 2*t.^2;
ulimit= 10*rand(size(x));
llimit= 10*rand(size(x));
hold on
h(1)=fill([t fliplr(t)],[x+3*ulimit fliplr(x-3*llimit)],'g','edgecolor','none')
h(2)=fill([t fliplr(t)],[x+2*ulimit fliplr(x-2*llimit)],'b','edgecolor','none')
h(3)=fill([t fliplr(t)],[x+ulimit fliplr(x-llimit)],'r','edgecolor','none')
set(h,'facealpha',.5)
plot(t,x,'b');
Kelly Kearney on 18 Sep 2015
My boundedline.m can help automate the procedure Joseph outlined:
% Data
t = (1:1000)';
y = 2*t.^2;
err = bsxfun(@times, rand(1000,12), t*1000);
y = bsxfun(@plus, y, err);
% Calculate bounds and plot
ymed = median(y, 2);
prc = [0 10 25 75 90 100];
p = prctile(y, prc, 2);
ylo = bsxfun(@minus, ymed, p(:,1:length(prc)/2));
yhi = bsxfun(@minus, p(:,end:-1:length(prc)/2+1), ymed);
bnd = permute(cat(3,ylo,yhi), [1 3 2]);
[hl, hp] = boundedline(t, [ymed ymed ymed], bnd);