Box Plots for Vectors of Varying Length

Hello! I have 3 data sets named DailyTMAX_1, DailyTMAX_2, and DailyTMAX_3. Each one of them have maximum temperature data. DailyTMAX_1 from years 1931-1960, DailyTMAX_2 from 1961-1990, and DailyTMAX_3 from 1991 to 2020.

The three of them have varying lenghts. DailyTMAX_1 is 10321, DailyTMAX_2 is 11243 and DailyTMAX_3 is 10792.

I want to create a boxplot but every time I run it I get this message:

"Error: File: ClimateDataProcessor.m Line: 107 Column: 14

Invalid text character. Check for unsupported symbol, invisible character, or pasting of non-ASCII characters."

Line 107 Column 14 corresponds to g1 on the attached code

 x = [DailyTMAX_1; DailyTMAX_2; DailyTMAX_3];
g1 = repmat({‘1931-1960’},10321,1);
g2 = repmat({‘1961-1990’},11243,1);
g3 = repmat({‘1991-2020’},10792,1);
g = [g1; g2; g3];
boxplot(x,g)

Hope anyone can help me with this

Thank you!

Answer 1

[Joel Lynch]

The proximate error is that the symbols around the year-range string

‘1931-1960’

are not apostrophes, so MATLAB doesn't know what to do with them. They should appear in pink:

'1931-1960'

But that is a moot point, because you don't need to use the "g" argument in boxplot for this problem. The g argument is used to group heterogenous elements of each column in x together in seperate boxplots. You're data is already homogenous in each vector, so you can just pass three columns in x to get three boxplots. The trick is to handle their unequal lengths, which you can do by filling a just-large-enough NaN matrix x. The extra elements in each column are ignored by boxplot.

N = max( [ numel(DailyTMAX_1), numel(DailyTMAX_2), numel(DailyTMAX_3) ] ); % Longest length
x = NaN(N,3); % 3 columns large enough to fit longest vector
x(1:numel(DailyTMAX_1),1) = DailyTMAX_1;
x(1:numel(DailyTMAX_2),2) = DailyTMAX_2;
x(1:numel(DailyTMAX_3),3) = DailyTMAX_3;
boxplot(x); 

You can greatly simplify this by storing "DailyTMAX_X" vectors natively in a NaN-padded x matrix.

Note: this won't add the year-labels, you can do that manually by:

set( gca,  'XTickLabel', {'1931-1960', '1961-1990', '1991-2020'} );

[Joel Lynch]

I just realized the method you used is straight from the MATLAB examples for boxplot. Your code should work for that after replacing apostrophes, but I personally think it's a very inelegant approach.

[Nuria Andreu]

Thank you Joel! I will try it out!

And yes, the method I used is straight from Matlab but I keep getting the same error message no matter how many times I try to fix it.

[Joel Lynch]

The symbol you are using appears to be the prime symbol, not the apostrophe; you'll want to check your system's keyboard layout to make sure you know which you're using.

One trick is to click on your matlab script, hit cntrl+h, and replace all ’ with '. That should remove all the bad characters from your script.

[Nuria Andreu]

Hi Joel thank you so much! It worked. The problem was the apostrophe. I didn't see it until you mentioned it.