Wrong area filled with fitlm confidence interval

Hello,

I have extracted the confidence interval from some data interpolated with the linear interpolation function in Matlab through the following commands:

mdl = fitlm(x_data, y_data);

Unrecognized function or variable 'x_data'.

x_data and y_data are column vectors with shape 43x1.

From which I have extracted the confidence interval:

[yhat1,ci1] = predict(mdl,x_data)

I want then to fill the area within the confidence interval and I have tried the following commands:

xconf = [x_data x_data(end:-1:1)];
yconf = [ci1(:,2) ci1(end:-1:1,1)];
yconf2 = [ci1(:,2) ci1(:,1)]

%First temptative
fill(xconf, yconf, 'red')
%Second temptative
fill(xconf, yconf2, 'red')

In both cases, the area which is filled is wrong (what is displayed is the confidence interval only):

I would like to ask some suggestions how to solve this problem having the confidence interval extracted with the linear interpolation function.

Thank you!

Answer 1

[Star Strider]

Use the patch function for best results —

x_data = linspace(-7, 5, 25).';
y_data = randn(size(x_data));

mdl = fitlm(x_data, y_data);

[yhat1,ci1] = predict(mdl,x_data);

figure
hold on
% plot(x_data, y_data)
plot(x_data, ci1, 'r')
patch([x_data; flip(x_data)], [ci1(:,1); flip(ci1(:,2))], '-r')
hold off

.

[Lucrezia]

Thank you! it works perfectly.

[Star Strider]

As always, my pleasure!