Plot on Boxplot on the same graph

Hello potential ansewrer,

I have let's say 9 differnent monthly matrices, days*hours (e.g. January 31*24).

Now I have calculated the hourly averages through the 31 days (=1*24 values) for each of the 4 matrices.

Then I am try to plot on the same graph, a boxplot of each month and by using hold on its hourly averages.

Of course, matlab does not accept this as monthly boxplots appear on position x=1 and its hourly averages appear x=1:24 so there are not consistent.

As the periods are the same I do not know if the use of multiple x-axis would help.

Boxplots:

Hourly Averages:

Any help :)

Answer 1

[dpb]

Shift/Rescale the hourly times to be centered about the 1:N values.

boxplot(T)
hold on
W=0.8;      % fractional width of each box -- arbitrary, adjust to suit/fit
for i=1:9
  t=rescale(1:24,i-W/2,i+W/2);      % normalize about month index
  plot(t,H(:,i));                   % plot hourlies against normalized times
end

[KonPol]

Ok, yes that could be a solution. I was wondering if I could something easier just by using hold on/plot/boxplot. What does T stand for?

[dpb]

What could be simpler than the above? It IS using hold on after creating the boxplot.

T is your array; create the boxplot first as shown from it and then add the other data.

Oh...I see, I changed horses in midstream -- the other T inside the loop should be H, the hourly data by month array instead of the boxplot data. My bad, didn't realize had done that.

Without having posted data or specific variables to work with, we're left just trying to make up generic stuff to fit...

[dpb]

Here's an example that the idea works with just arbitrary data --

boxplot(randn(30,9))        % a boxplot of 30 "days" by 9 months
hold on
t=1:24;                     % the 24 hours vector
W=0.8;
t1=rescale(t,-W/2,+W/2);   % rescale around zero as base
for i=1:9                   % over each month group
  plot(i+t1,randn(24,1),':')    % add a centered graph over each bar in turn
end
xticklabels(month(datetime(2022,1:9,1),'shortname'))

Just use your data in place...

Looks like W could/should be a little smaller for the nine boxes; the range of the added lines is somewhat wider than the boxes themselves. You'll have to 'spearmint and see what is proper adjustment.