Making a box plot

I have a 10x51 matrix and was trying to make 2 box plots with the first 5 rows showing column 1 data and the next 5 rows of showing column 1 data as well. For example if the matrix had 10 numbers with the first 5 being 1 and the next 5 being 2. How do I make a figure that creates two box plots and compares them against each other. Need to do this for all 51 columns of data.

Answer 1

[Adam Danz]

I'm not sure how you'd like to arrange the 51 plots.

Here's a template you can follow. It plots the first 6 columns data but you could change that by changing 'n' to whatever you'd like. Just keep in mind that the figure will become cluttered very quickly.

I first create fake data then I loop through the first 'n' columns and compare the first and second halves of the column as you described.

%fake data
m = rand([10,51]) .* randi(50, 1,51); 

figure
n = 6; 
for i = 1:n
    subplot(1,n,i)
    tempData = reshape(m(:,i), [], 2); 
    boxplot(tempData)
    xlabel('col half')
    title(sprintf('Col %d', i)); 
end

Here's the version that produces a new figure every 'n' columns so that all of your data are plotted.

% How many subplots per figure?
n = 10; 

% That will produced this many figures
nFigs = ceil(size(m,2) / n); 

% Loop through figures
c = 0; %column counter
for j = 1:nFigs
    figure
    for i = 1:n
        c = c+1; 
        if c > size(m,2)
            continue
        end
        subplot(1,n,i)
        tempData = reshape(m(:,c), [], 2);
        boxplot(tempData)
        xlabel('col half')
        title(sprintf('Col %d', c));
    end
end

[Jay Stewart]

I copied your code and only got 1 box plot to show up. I changed the n from 6 to 10, hoping to get 10 plots but that didn't work. I was also wondering as to how to create another figure with the next 10 columns as I don't want to reuse any data from the previous columns

[Adam Danz]

You must have copied it wrong or maybe you didn't execute the entire code. I'm assuming you copied and executed my fake data as well.

I just copied the whole thing and ran it again and it works (produced the same image but with different random fake data).

I updated my answer to show you how to produce a new figure every 10 columns. Tested, works.

If it doesn't work with your data we need to find out how your data differ from my fake data.

[Jay Stewart]

Okay, it works perfectly. I have an additional question because instead of sorting the column by 5 and 5, I need to label certain rows as x and other rows as y and then plot them against each other. This would organize them by class, essentially. How would I do so?

[Adam Danz]

I'm glad it worked! If you accept the answer (by pressing accept) future viewers will know that this solution worked.

What have you tried so far to implement your new idea?

I'll give you a hint

help boxplot

Read about "boxplot(X,G)"

[Jay Stewart]

I made a matrix created with 1s and 0s to distinguish between the data to make each box plot. I tried to add it to the line of code you made simply with boxplot(tempdata,g), but that did not render all the box plots. only about half. what should I try next?

[Adam Danz]

In this example, I make two groups of data (groups 1 and 2) and I assign each row of the matrix 'm' to a group.

groups = [1 1 2 2 1 1 2 2 1 2]'; 
for i = 1:n
    subplot(1,n,i)
    boxplot(m(:,i), groups)
    xlabel('col half')
    title(sprintf('Col %d', i)); 
end

[Jay Stewart]

Okay, that fixed the problem of not having all the box plots showing up and classifying them by their respective groups. However, now the same box plot is rendered for all 51 columns

[Jay Stewart]

This is the current code

if true
  % code
end
n=2;
nFigs = ceil(size(m,2) / n); 
c=0;
for j = 1:nFigs
  figure
  for i = 1:n
      c = c+1; 
      if c > size(m,2)
          continue
      end
      subplot(1,n,i)
      boxplot(m(:,1),groups)
      xlabel('col half')
      title(sprintf('Col %d', c));
  end
end

[Adam Danz]

That's because you're only requesting the first column of M to be plotted on each figure.

boxplot(m(:,1),groups);  % m(:,1) is the first col of m

replace 1 with c to iterate through each col of m.

[Jay Stewart]

But then I got the error: "Index in position 2 is invalid. Array indices must be positive integers or logical values

[Jay Stewart]

because c is just a 1x1 double matrix with its lone value being 52

[Adam Danz]

Then something else is wrong in your code. c should be a double, 1x1 matrix and as your code indicates, iterate through 1 to 51.

If I copy your code from 4 comments up, replace the '1' in the boxplot() call with c, and I add the grouping variable

groups = [1 1 2 2 1 1 2 2 1 2]';

it works.

[Adam Danz]

Here's your code with the two changes I summarized above. Copy this and run this on your matrix 'm'. If you have an error, please copy the full error message into the comments section and copy the line of code that produced the error.

groups = [1 1 2 2 1 1 2 2 1 2]';   % Add this
n=2;
nFigs = ceil(size(m,2) / n); 
c=0;
for j = 1:nFigs
  figure
  for i = 1:n
      c = c+1; 
      if c > size(m,2)
          continue
      end
      subplot(1,n,i)
      boxplot(m(:,c),groups)       %Changed 1 to c
      xlabel('col half')
      title(sprintf('Col %d', c));
  end
end

Answer 2

[Jay Stewart]

I don't understand what the problem could be. I re ran the code again. Same issue