How to plot based on the shortest vector?

I want to plot two collumns of data against each other but one is longer than the other. The data all lines up and there is only a couple of extra data points at the bottom of the collumn.

Here is the code in question:

patch(po,pa,x,'Facecolor',[0.8,0.8,0.8]);

hold on

AltvO3=plot(O3ave,adj_ave_alt,'-o','Markersize',3,'color','k');

In this case I want pa and x to include up to as many data points in po and adj_ave_alt to include up to as many data points as O3ave.

[Rik]

The easiest solution is to write a wrapper function that trims the data as you need and then calls plot or patch with the shortened arrays.

[Joseph Tirado]

I think you are overestimating my abilities. How would I do that?

Answer 1

[Scott MacKenzie]

So, the approach I think you are willing to take is just to ignore the few extra elements that are in the longer of the two vectors. Here's a simple way to do this:

% two column vectors of data with a few extra elements in one of them
c1 = randi([1 100], 20, 1);
c2 = randi([1 100], 18, 1);

% determine the length of the shorter of the two columns of data
n = min(numel(c1), numel(c2));

% plot the data, ignoring the extra elements in the longer vector
patch(c1(1:n), c2(1:n), 'r');

[Rik]

You should use numel or size. I have not yet seen any situation in which length is a better choice.

[Scott MacKenzie]

@Rik Noted. This just a matter of execution time, right? The difference must be miniscule for vectors, or am I missing something?

[Rik]

I have not tried measuring the execution time, but the differences should be minimal or non-existent.

The broader point is expectations.

If the input array is a vector, there is no difference in output between length and numel. But what kind of errors will you get you suddenly get an array input? If you use it to generate loop indices, you will suddenly only process part of the array. You might get unexpected behavior much later in your code.

If you explicitly tell Matlab what to do, there is no opportunity for confusion. You either want the number of elements, or you want the size along a specific dimension. There will never be a bug or unexpected result.

Why would you want max(size(A)) if you can actually specify what you mean? You say the result is the same for vector inputs, but your code doesn't test if the input actually is a vector.

[Rik]

I got a notification you edited your answer, but since you didn't replace length with numel, I could really find a material edit or a response to my comments.

[Scott MacKenzie]

@Rik I can't remember but I probably just edited the code comments or fixed some typo in my text. You're right in that I did not at that time change length to numel. I just made the change.

I'm not convinced this is a major issue. Personnaly, I tend to use length with vectors and don't see any imperative in using numel. But, I get your point that people may sometimes have incorrect expectations.

[Rik]

I just don't see the merit of using something that has the potential to cause issues, while there is a function that does the exact same thing for vectors, but will not cause unintended behavior for array inputs.

I don't see why you would teach anyone a tool that may bite them, while there are viable alternatives. Can you convince me why numel would be a worse choice?

[Scott MacKenzie]

This seems like splitting hairs to me. If you think this is significant, perhaps take it up with MATLAB staff. They could easily add "not recommended" in the documentation, as done with other functions people should avoid.

Answer 2

[Rik]

I was just about to comment something along the lines of the answer @Scott MacKenzie posted. The idea is that you find out what the shortest vector is and then call the function you need. The only extra part I was suggesting was a function definition and some documentation:

h=short_plot(rand(10,1),rand(8,1))

h = 
  Line with properties:

              Color: [0 0.4470 0.7410]
          LineStyle: '-'
          LineWidth: 0.5000
             Marker: 'none'
         MarkerSize: 6
    MarkerFaceColor: 'none'
              XData: [0.9257 0.3860 0.4225 0.3850 0.0851 0.7865 0.0147 0.9023]
              YData: [0.4146 0.5354 0.7882 0.1400 0.9720 0.7587 0.7900 0.7639]
              ZData: [1×0 double]

  Show all properties

function varargout=short_plot(x,y,varargin)
%This functions shortens either x or y to make them equal length and calls
%the plot function. It will forward all optional inputs to plot().
n=min(numel(x),numel(y));
x=x(1:n);y=y(1:n);
varargout=cell(nargout,1);
[varargout{:}]=plot(x,y,varargin{:});
end

You can do the same for your custom patch function.