Aslak Grinsted

Nice, useful and compact code! It would be nice if i could give it a list of handles to put in the legend.

@mohammad: The code does do MLR. Here's an example with multiple predictors:

X=randn(100,4);
y=X*[4 2 4 1]'+randn(100,1)*.2;
[p,stats]=quantreg(X,y,.9);

I slightly modified Matthias code below to make it work for an arbitrary long horizontal subregion subplot(m,n,x:y), e.g. subplot(2,3,1:3) or subplot(3,3,4:9):

if numel(Args.NumericArguments{1}) == 2 % restore subplot(m,n,[x y]) behaviour
[x1 y1] = ind2sub([Args.cols Args.rows],Args.NumericArguments{1}(1)); % subplot and ind2sub count differently (column instead of row first) --> switch cols/rows
[x2 y2] = ind2sub([Args.cols Args.rows],Args.NumericArguments{1}(end));
else
x1=mod((Args.NumericArguments{1}-1),Args.cols)+1; x2=x1;
y1=floor((Args.NumericArguments{1}-1)/Args.cols)+1; y2=y1;
end

I could restore the subplot(2,2,[1 3]) behaviour by changing the case 1 in the switch length(Args.NumericArguments) block (line 64) to this:

if numel(Args.NumericArguments{1}) == 2 % restore subplot(m,n,[x y]) behaviour
[x1 y1] = ind2sub([Args.cols Args.rows],Args.NumericArguments{1}(1)); % subplot and ind2sub count differently (column instead of row first) --> switch cols/rows
[x2 y2] = ind2sub([Args.cols Args.rows],Args.NumericArguments{1}(2));
else
x1=mod((Args.NumericArguments{1}-1),Args.cols)+1; x2=x1;
y1=floor((Args.NumericArguments{1}-1)/Args.cols)+1; y2=y1;
end