How to use boxchart()?

I am trying to use a boxchart for the first time as opposed to a boxplot, as it looks to have better functionality for what I want to do.

But, it's not working and I can't understand what's wrong...

On the x axis I would like to have 22 different dates, represented on the y by a box plot of 22*144 data points.

boxchart(data_dB1,dates_concatenated)
Error using boxchart
Expected ydata to be one of these types:

double, single, uint8, uint16, uint32, uint64, int8, int16, int32, int64

Error in boxchart (line 95)
validateattributes(ydata,{'numeric'},{'2d','real'},mfilename,'ydata');

Answer 1

[Cris LaPierre]

I think the error is because you are using the syntax boxchart(xgroupdata,ydata), but your inputs are reversed. You put the grouping variable second. Also, your grouping data is not valid for use as xgroupdata.

From the documentation, if you want to use a cell array of characters, you must use the syntax

boxchart(___,'GroupByColor',cgroupdata)

Therefore, try changing your code to

boxchart(data_dB1,'GroupByColor',dates_concatenated)

[Louise Wilson]

Thanks Chris. I am sorry, I see now I was getting it wrong and my question wasn't as clear as it could be. I realise that I do not want to group by date, but by three different categories. I can do this:

h=figure(7);
subplot(5,1,1)
set(h,'WindowStyle','docked')
boxchart(data_dB1);
hold on
boxchart(data_dB2);
boxchart(data_dB3);
ylim([80 150]);
xticklabels(datestr(dates_concatenated1,'dd-mm-yy'));
xtickangle(45);
legend(["50-24,000 Hz","50-5000Hz","50-1500 Hz"]);
title(site,'FontSize',14);
xlabel('Date','FontSize',14); ylabel('SPL_{rms}', 'FontSize',14);

...which will plot a boxchart for each different category on top of each other within one xtick. But, what I want to do for clarity is to have three boxcharts side-by-side within one x-tick, so they can be seen clearly. So, the question is, how do I group these data correctly? Each 'group' is 144*22 double matrix and belongs to one of three categories.

[Cris LaPierre]

With hold on, it is doing exactly what it should - placing the plots on top of each other. You will need to include all the data in a single boxchart command.

If you have the Statistics and Machine Learning toolbox, you can also try using boxplot.

[Louise Wilson]

Yes, I understand it is doing as it should, just that what it should do is not my ideal solution. I have tried to use boxchart as it has legend whereas boxplot does not. I have concatenated all of my data into two arrays:

C=cell(432,22); %labels for each category
C(1:144,1:22)={'50-24,000'};
C(145:288,1:22)={'50:5000'};
C(289:432,1:22)={'50:1500'};

dB_values=[dB1; dB2; dB3]

b=boxchart(dB_values,'GroupByColor',C)

but I get the error:

Error using matlab.graphics.chart.primitive.BoxChart

The name 'GroupByColor' is not an accessible property for

an instance of class

'matlab.graphics.chart.primitive.BoxChart'.

Error in boxchart (line 128)

H = matlab.graphics.chart.primitive.BoxChart('Parent',

cax,...

From the example in the documentation using the patients data, this is what I should do, but it is not working?

[Cris LaPierre]

Share your data. Place your variables in a mat file and attach them using the paperclip icon.

[Louise Wilson]

Thanks

[Cris LaPierre]

What are data_dB1, data_dB2, and data_dB3?

[Louise Wilson]

They are here, and concatenated vertically in dB_values. They are dB values.

[Cris LaPierre]

I had to do some investigation myself first. I found this example extremely helpful. Basically, in order to create the plot you want using boxchart, you need to convert your inputs into vectors. All the data needs to be in a single column. To work, you also need to do the following

• create an xgroup vector that indicates which column the corresponding value originally came from.
• create a cgroupdata vector that indicates which data set the corresponding value came from.

I came up with something like this.

% load your data
load data_dB1.mat
load data_dB2.mat
load data_dB3.mat

% Convert each data set into a column vector.
% the (:) basically stacks all columns on top of each other. First #1, the 2 under it, then 3, ...
dB1 = data_dB1(:);
dB2 = data_dB2(:);
dB3 = data_dB3(:);

% Combine all 3 data sets into a single column vector
dB = [dB1;dB2;dB3];

% Create a corresponding x-vector, indicating which column each row came from
x=ones(size(data_dB1,1),1)*(1:size(data_dB1,2));
x = [x(:);x(:);x(:)];
% convert to dates
d=datetime(2019,8,20):days(1):datetime(2019,9,10);
d=categorical(d(x));

% Create a group vector, indicating which data set each row came from
G = ones(size(dB1));
G = [G;G*2;G*3];

% Create boxchart
boxchart(d,dB,'GroupByColor',G)

ylim([80 150]);
legend(["50-24,000 Hz","50-5000Hz","50-1500 Hz"]);
xlabel('Date','FontSize',14); ylabel('SPL_{rms}', 'FontSize',14);

[Louise Wilson]

Hi Chris, this is fantastic. Thank you so much for your help, this is exactly what I was looking for. I hadn't seen that example either.

[Ioannis Vourvachakis]

Hello. How can I use boxchart? My version of Matlab is 2019a ans sais Undefined function or variable 'boxchart'.

[Cris LaPierre]

boxchart was introduced in R2020a. If you have the Statistics and Machine Learning toolbox, you can use boxplot instead. Otherwise, you will need to update your MATLAB version to at least R2020a.

[Fuqiang Guo]

Thank you very much, your advice is very helpful, I'll update my software version