File Exchange

image thumbnail

Box and whiskers plot (without statistics toolbox)

version 1.0 (13.9 KB) by

Creates nice boxplots from data. You don't need a toolbox. Simple yet fully featured.

4.78571
17 Ratings

135 Downloads

Updated

View License

bplot(X)
This function will create a nice boxplot from a set of data. You don't
need a toolbox.
bplot(D) will create a boxplot of data D, no fuss.
T = bplot(D) If X is a matrix, there is one box per column; if X is a
vector, there is just one box. On each box, the central
mark is the median, the edges of the box are the 25th and
75th percentiles
array 'T' for a legend. You can add the legend as legend(T)

T = bplot(D,x) will plot the boxplot of data D above the 'x' value x

T = bplot(D,y,'horiz') will plot a horizontal boxplot at the 'y' value of y

T = bplot(...,'Property', . . . )
T = bplot(...,'PropertyName',PropertyValue, . . . )

% Jitter feature
The boxplot has a cool jitter feature which will help you view each
outlier separately even if two have identical values. It jitters the
points around the other axis so that you can see exactly how many are
there.

% Examples:
bplot(randn(30,3),'outliers')
bplot(randn(30,3),'color','black');
----
X = round(randn(30,4)*5)/5; % random, with some duplicates
T = bplot(X,'points');
legend(T,'location','eastoutside');

% This function was created for inclusion in my larger nhist function.

Comments and Ratings (25)

hello

hello (view profile)

Suppose I have a matrix D which is 3 by 4. So that means I will have four box plots. How can I label x-axis for each box plot?
For example I would like to label A for first box plot, B for second and so on.

Adam

Adam (view profile)

If you comment out lines 320-331 the function works fine. Something must have changed with the properties of rectangles or legends.

I am unsure if Jonathan Lansey still checks this page; the last update was Sep 2015.

I have nhist.m in one of my public repos at https://github.com/erikrtn/dataviz/. GitHub is a better tool for collaborating on and sharing code vs. File Exchange.

Perhaps we can work out of there. Send pull requests and let's keep improving this code!

Subsequent versions of this code should attribute Jonathan for his great work, and link to this File Exchange page.

Jonathan, if you check this and are unpleased and would prefer to host this function in one of your own GitHub repos, we should do that. Let us know. Thanks.

betagamizado

It did not work for me. I used a simple matrix:

A = [1 2 3 4 5;
2 3 4 5 6;
1 2 3 4 5;
4 2 4 5 2;
3 5 6 7 7]

bplot(A)

%but I get the following error message:

Error using get
There is no Annotation property on the Rectangle class.

Error in bplot (line 329)
set(get(get(hReg2(ii),'Annotation'),'LegendInformation'),'IconDisplayStyle','off'); % Exclude line from legend

Error in bplot (line 97)
bplot(x(:,ii),ii,'nolegend');

% also, this yields only two boxplots, but it was supposed to yield five, since the input has five columns. I don't know how to fix that.

I'm using matlab 2016a

Michael Zint

Very useful tool!
However using Matlab 2016b it seems that user defined xlabels are not supported any longer. xlabels always defined from 1:n. Entiers for x seem to be ignored. In older versions the xlabels are plotted as specified in vector x.
Does anyone have an idea on how to fix this.

Siwei LOU

After looking at documentation on box and whisker plots, I am not certain where you got that the wisEdge variable (essentially accounting for outliers or not) is 9% on either side. Currently line 254 looks like:
wisEdge = prctile(x,[percentileNum2 100-percentileNum2])

I would change it to:
wisEdge = [boxEdge(1)-IQR*1.5 boxEdge(2)+IQR*1.5];

as is conventional with many learning resources.

When trying to use 'box' option for multiple distributions on the same figure, I get this error:

===

Error using get
There is no Annotation property on the Rectangle class.

Error in nhist>bplot (line 2073)
set(get(get(hReg2(ii),'Annotation'),'LegendInformation'),'IconDisplayStyle','off');
% Exclude line from legend

Error in nhist (line 1169)
bplot(logFunc(cellValues{k}),tempY,'color',lineStyleOrder{k},'barwidth',boxplotWidth,'linewidth',linewidth,'horizontal',pointsOrNo,'specialwidth','histmode');

===

I solved it by simply commenting out lines 2070 to 2075:

% % remove all remenants of legends
% if forceNoLegend
% for ii=1:length(hReg2)
% set(get(get(hReg2(ii),'Annotation'),'LegendInformation'),'IconDisplayStyle','off'); % Exclude line from legend
% end
% end

Seb Biass

Update on Maider's comment, replace lines 322-331 with Maider's suggestion

Hugo

Hugo (view profile)

There is a minor bug in the code, where feeding an all-nan array raises an error instead of being treated as an empty array. The fix is quite simple, move the lines in the section 'check that there is at least some data' after the section 'serialize and remove NANs'.

Hugo

Hugo (view profile)

perhaps the documentation could make more clear all NaN's are actually filtered out, this helps for plotting potentialy unequal sized columns next to each other.

Tim

Tim (view profile)

idontknow

Maider

Maider (view profile)

I found the same error than Brian and DKM I found a way to solve it by changing lines 304-313 to

for ii=1:length(hReg)
set(hReg(ii),'HandleVisibility','off')
end

% % remove all remenants of legends
if forceNoLegend
for ii=1:length(hReg2)
set(hReg2(ii),'HandleVisibility','off')
end
end

Maider

Maider (view profile)

Thank you very much for submitting this code. Works well, easy to customize and save me ton of time :D

DKM

DKM (view profile)

I've got the same issue as Brian, "no Annotation property on the Rectangle class" error. Worked fine until I upgraded to 2015a. I wonder what changed?

chsharrison

Nice contribution, thank you. Really like that I can overplot box plots on top of other plots, Matlab's boxplot() will not do this.

Brian

Brian (view profile)

When I try to plot a matrix D, larger than a vector, in MATLAB R2015a, I get the following error: (I don't get the error in R2014b)

"Error using get
There is no Annotation property on the Rectangle class.

Error in bplot (line 311)
set(get(get(hReg2(ii),'Annotation'),'LegendInformation'),'IconDisplayStyle','off'); % Exclude line from legend

Error in bplot (line 99)
bplot(x(:,ii),ii,'nolegend');"

When plotting a vector such as: "bplot(z(:,1));" It plots fine, but as soon as we go to a higher dimension (such as: "bplot(z(:,1:2));") I receive the above error.

F Z

F Z (view profile)

It's very useful. Is there any way to display values on every plot please?

Gavin

Gavin (view profile)

Easy to produce plots that communicate results well

Gavin

Gavin (view profile)

Tim

Tim (view profile)

Exactly what I've been looking for. Love the simplicity of the features!

Udit Gupta

It's quite useful. Wish there was a way to perform the analysis when the number of observations for each experiment are not the same.

Updates

1.0

Minor update

MATLAB Release
MATLAB 7.11 (R2010b)

Download apps, toolboxes, and other File Exchange content using Add-On Explorer in MATLAB.

» Watch video

Win prizes and improve your MATLAB skills

Play today