Home –  Archive
Monthly Archives: Jan 2015

scaleSubplot, to organize your subplot nicely

Download scaleSubplot

In psychology we often deal with data from several subjects, each subjet having several datapoints across more than one condition. Showing the results of an experiment for individual subjects in order to get a meaningful insight understandment of the data could be quite tricky. A classic way of organizing the data consists of plotting each subject's result in a subplot, in order to have all the subject's data in a single figure. MATLAB function subplot is quite good for the task, even if more and more often I find myself using the tight_subplot function. Anyway, one problem is that the subplot axes does not have the same scale by default, and this decrease the redeability of the results.

MATLAB allows you to link different axes (namely different subplots) so that you can scale all of them at the same time (linkaxes).  However, once that all the subplots have the same scale, showing the axes label for each subplot seems a little bit redundant. It would be sufficient just to show the axes of the bottom and left most panels. I coded a function that does exactly that.

scaleSubplot(fig, varargin) allows you to scale at the same time all the subplots of a figure. You can specify a custom scale or let the code decide a nice unique scale for all the subplots. You can decide to scale at the same time the horizontal and vertical axes, or only one of them.  Let's look at some example.

The easiest way of using the function is: scaleSubplot(gcf). In this case you will scale both x and y axes, and the script will find the best xlimit and ylimit for all the subplots. The code will make sure that all the data will be visualized in the subplots.

Note how the redundant x/ylabels have been suppressed:

Untitled

We can have more control on the image by specifying some optional arguments. For example, if we want to have the same scale only the x axis, we specify 'sameScale',{'x'}. Notice how the x labels on the intermediate subplots are suppressed, but all the y labels are still there, since they do not have the same scale. In the figure below we also specify a custom xlimit. If you decide to specify a custom limit, make sure that all the datapoints in all the subplot fall within you limits.

Untitled2

We can also specify only one of the boundaries for the axes limit, and leave the code decide for the other one. For example we can write scaleSubplot(gcf,'xlimit',[-Inf 40]) or scaleSubplot(gcf,'ylimit',[0 Inf]).

My function is expecially useful when we want to have a tight subplot. In this case, eliminating the intermediate axis can make the subplot much more readable. For example, I generate the image on the left side with tight_subplot(2,4,0.05), where 0.05 is the gap between subplots. However, with all the axes labels, the image is a mess. After a simple scaleSubplot(gcf) the figure looks much better.

Untitled3

Hope you enjoy it.

Download scaleSubplot