How can I make a boxplot match at 0 with another plot?

Hi

I'm having trouble making a boxplot match at the origin with another plot (colored lines). The first plot works just fine, but as the boxplot works with a categorical x axis, it shifts away from the actual origin and I just cant make it match back into place.

Can anyone help me, please.

[the cyclist]

Here are a few things that would make it easier to help you. If you can do some or all of these things, you might have a better chance at getting a solution:

• Upload your data
• Upload your code
• Make a figure with just two panels: One that is how you want it, and one that is not how you want it
• Explain what you mean by "it shifts away from the actual origin". I'm not sure what "it" is (e.g. are you talking about the boxes, or the lines, or something else?). I don't know what the "actual origin" means.

Mainly, I don't know what exactly is supposed to "match at the origin". And even if I did understand that, it's probably not possible to give advice about how to fix that, without seeing your code, your data, or both.

Answer 1

[Constantino Carlos Reyes-Aldasoro]

A workaround to solve your problem is to call the boxplots with zero values in between your actual data. E.g.

plot([1 2 3 4 5 6 7],[ 9 8 7 6 5 4 3 ],'-o')
hold on
x1 = rand(50,1);
x2 = ones(50,1);
boxplot([0 ; 5*x1 ; 0 ;0; x1*4;0;x1*8],[1; 2*x2; 3 ;4; 5*x2;6;7*x2])

In this way, the boxplots are located exactly at the same location as the other plot.

[the cyclist]

If this approach solves the issue, I would suggest filliing in with NaN rather than zeros. Then no box at all appears.

plot([1 2 3 4 5 6 7],[ 9 8 7 6 5 4 3 ],'-o')
hold on
x1 = rand(50,1);
x2 = ones(50,1);
boxplot([NaN ; 5*x1 ; NaN ;NaN; x1*4; NaN; x1*8],[1; 2*x2; 3 ;4; 5*x2;6;7*x2])

[Constantino Carlos Reyes-Aldasoro]

that is better!

[Isaac Elías]

Thak you both!

This advice helped to make the boxes match my actual data for times above 0. Although I could´nt make the first box appear at time zero of the line plot (wich apparently, is not possible, as boxplot works with categorical data).