raacampbell/shadedErrorBar
- 15 files
- 536 downloads
Editor's Note: This file was selected as MATLAB Central Pick of the Week
shadedErrorBar is a MATLAB function that creates a continuous shaded error region around a line rather than discrete bars. The error region can either be specified explicitly or calculated on the fly based upon function handles. The appearance of the plot can be modified via input argumets and/or the handles of the returned plot objects. You will need OpenGL to get transparency to work.
There are detailed examples along with example images on the GitHub page: https://github.com/raacampbell/shadedErrorBar
Comments and Ratings (128)
Updates
| 1.65.0.0 | Provide "patchSaturation" optional input argument. Update demos. |
|
| 1.62.0.0 | stars |
|
| 1.62.0.0 | github for questions |
|
| 1.62.0.0 | update description |
|
| 1.61.0.0 | move to GitHub |
|
| 1.6.0.0 | Verified to work with R2015a |
|
| 1.5.0.0 | Handles now only returned if the user requests them. Plot elements are re-ordered in a more elegant manner. |
|
| 1.4.0.0 | Improved error-checking of input args and revised some of the comments. |
|
| 1.3.0.0 | Correct bug that caused patch area not to be plotted when y contained NaNs and error bars were generated using the function handles approach. |
MATLAB Release Compatibility
Platform Compatibility
Windows macOS LinuxTags
Acknowledgements
Inspired: superbar, shplot, errorshade, Murphy Diagrams, shade_significant_timepoints, fillPlot(mat,varargin), boundedline.m, Mean square displacement analysis of particles trajectories, Shaded Error Bar YY, mseb(x,y,errBar,lineProps,transparent)
Discover Live Editor
Create scripts with code, output, and formatted text in a single executable document.
ibrahim kurban ozaslan (view profile)
Han Zerui (view profile)
It's awesome! thx!
Rob Campbell (view profile)
I think I would just DIY the legend. Make some patches and some labels manually. Maybe there's a better way, but I don't know what it is. More likely to look nice if you roll your own, though.
xiaolong wang (view profile)
Great work.
I have an issue. when I plot the error bar. The color of the legend is different from the color of the plot. It is because of the property of the patch. The patchSaturation is set to 0.2. (It is necessary then we can see several error bar).
how do we deal with the difference of color? which is important when we plot.
thanks a lot.
Chao Gu (view profile)
Great work!!!
xiaochuan sun (view profile)
Javed Niazi (view profile)
How to use this function for plotting multiple plots with respective legends using it in a for loop?
Rob Campbell (view profile)
Robert, I would just interpolate through the NaNs then feed these new data to the function. I don't follow your other question, though. The last example doesn't involve NaNs and for me it works. Can you explain how it doesn't work?
Robert (view profile)
Very nice and helpful function! Wondering if would be possible not to plot NaNs values (e.g. middle or extremes of a vector)? I followed your last example (https://github.com/raacampbell/shadedErrorBar), but cannot make it work. Any more specific suggestions on how to do this will be very much appreciated!
eva zita patai (view profile)
Rashid Al Mukaddim (view profile)
Abbey Roelofs (view profile)
Joshua Baker (view profile)
Has anybody managed to create a 2D error plot, as in, X x Y with shaded deviations on each axis?. e.g. i am people attempts at drawing circles on a graphics tablet. I have both X and Y vectors, each with std across the board. I was to plot X x Y with shaded error on both axes. Thank you
Thierry Kremeyer (view profile)
Has anyone found a way to keep the shading when using a semilog plot with
set(gca,'YScale','log') ?
Seems like something is wrong with the patching?!
András Pálffy (view profile)
Anyone who wants to plot RGB defined colors, can do it like this:
'lineprops', {'color', [0 0 1]}, which plots a blue line, and so on.
Rob Campbell (view profile)
If you can supply it as a function handle, it should work.
Lief Fenno (view profile)
Hello! This is great. Quick question: do you have a version or know of a quick change so that I can pass a curvefit/continuous equation to the parser and use the 95% confidence intervals as the error bars? Thank you :-)
Rob Campbell (view profile)
You can set those values to NaNs via the returned plot object. See last example: https://github.com/raacampbell/shadedErrorBar
Gonzalo Lerner (view profile)
Thanks! Really useful. I have a problem though. I have a Vector that contains nans in the middle. I would like the function not to plot the errorbar on those NAN values. Is there a way to do it?
Jan Cagan (view profile)
Alex Henderson (view profile)
Wow, that's a top tip!
Rob Campbell (view profile)
You need to save it as a vector file. Change the figure window "Renderers" property to "painters" and re-save it.
Britt Veldman (view profile)
This is a very helpful function, thanks!
I saved the figure as an .eps file and am trying to edit it in Adobe Illustrator, but cannot access the elements like usually with .eps files. Do you have any suggestion as to how I can do this? Thanks.
prasad (view profile)
Henrik (view profile)
Gregory Canal (view profile)
Rob Campbell (view profile)
That's probably a mistake that crept in when I was testing something. It'll show up only if you try asymmetric error bars. I corrected it and added an example in the demo showing how the asymmetric error bars work.
Gregory Canal (view profile)
I'm comparing to an older version of shadedErrorBar, and it seems the error bar calculation has been switched from
uE=y+errBar(1,:);
lE=y-errBar(2,:);
to
uE=y-errBar(1,:);
lE=y+errBar(2,:);
Was the definition of errBar negated in some release?
Ian Kim (view profile)
This. Is. Great. Thank you
Rob Campbell (view profile)
Yes, you can edit the plot objects after they are rendered. See the examples and here: https://github.com/raacampbell/shadedErrorBar and also for more general info here: https://ch.mathworks.com/matlabcentral/answers/151011-how-to-plot-a-line-of-a-certian-color
Mehdi (view profile)
Very helpful function. I wonder if 'lineProps' command only works with standard colors defined in Matlab (i.e. m,k, etc)? Is there a way to specify color outside of the base with user defined [R G B] style?
Valentina Unakafova (view profile)
Rob Campbell (view profile)
William, what about rotating the axes? https://ch.mathworks.com/matlabcentral/answers/228449-how-to-rotate-plot-using-code
Rob Campbell (view profile)
America. I will see if I have time next week to look into that. Never tried it with yy axis. Can you also either e-mail me a minimal example for producing that error (see my profile for contact details) or add the info here: https://github.com/raacampbell/shadedErrorBar/issues/12 I prefer not to these comments as this isn't a good place to handle support issues.
Ame ZL (view profile)
Hey, thanks for the function, it works beautifly when you have only one y axis.
However, i'm trying to use it with two graphs using yyaxis and it's sending an error message, any clue on how can i fix it? thanks!
Error using matlab.graphics.axis.Axes/set
Children may only be set to a permutation of itself
Error in uistack (line 134)
set(UParent,'Children',AllChildren);
Error in shadedErrorBar>makePlot (line 193)
uistack(H.mainLine,'top') % Bring the main line to the top
Error in shadedErrorBar (line 126)
H = makePlot(x,y,errBar,lineProps,transparent,patchSaturation);
William J Riley (view profile)
I'd also like to make a horizontal shaded error plot. I see you discussed this in Nov (22, 24). Is there a way to do this with this tool?
Sébastien Herbert (view profile)
Maksym Sich (view profile)
Lionel Arteaga (view profile)
Hi all. Is it possible to make invisible the shaded error bar when "errBar" is NaN? As it is right now, it seems to be linearly interpolating the shaded area from the last non-nan to the next one. Thanks!
Saivimal Sridar (view profile)
f (view profile)
Rob Campbell (view profile)
Matteo, can you give me an example of what you're trying to do? Do you just want to rotate the axes?
Matteo Lisi (view profile)
nice solution but how can i plot 'horizontal' errobar?
Adam Danz (view profile)
Very nice tool. In case anyone else was unclear, when 'errBar' is a vector, the shaded error is drawn symmetrically as +/- errBar rather than centering the error about y. This is demonstrated by: shadedErrorBar([1:.1:10],repmat(3,1,91), repmat(2,1,91)); ylim([0,6])
It would be nice to have the ability to plot error in the x axis rather than Y.
Rita (view profile)
Thanks! I have problem with legend of two plots that I used hold on for them but the legend just shown for the last plot.
Dirk Isbrandt (view profile)
Fabrizio Schiano (view profile)
Thanks. Really appreciated!
Charles Lu (view profile)
Krithi Gopalan (view profile)
This function is very useful, thank you so much! My only issue is that I cannot plot multiple shadedErrorBars in app designer. It will pop up in another figure window and I can't figure out how to hold on using the function. Could someone help with this? Thanks!
Rob Campbell (view profile)
Aditya - the code doesn't do that right now. It's a good suggestion. I'll implement it when I have time.
Rob Campbell (view profile)
Istvan, did you try the last input argument that handles transparency?
Joe (view profile)
Istvan Kiss (view profile)
Thanks for this but can you let me know how can I plot in such a way that two plots do not completely cover each other but rather come out as on your example?
Guillermo Serrano Nájera (view profile)
Thanks a lot for this function!!
Teresa Serradas Duarte (view profile)
Thank you for the function! Is there any way to save the figure in such a way that it enables changing it in illustrator? saveas(figure,path,'epsc') doesn't work...
Vinil T C (view profile)
Great function!
Guy Bouvier (view profile)
cristina rigato (view profile)
Very useful function, thanks a lot!! It saved me much time and lines of code :)
kelian dascher-cousineau (view profile)
Aditya (view profile)
Can you please provide a working example where we have asymmetric error bars?
Dimitrios Karagkiozidis (view profile)
Antonin Blot (view profile)
Julian Rüdiger (view profile)
Thanks for this great code!
I got some troubles using datenum values on the x-axis. I have data points every 0.5 seconds over 30 minutes, but I want to plot the actual time on x instead just a duration. With the duration from 0 to 30 minutes I got no problems, but with the datenum-values the error shades form triangular areas all over the plot. Did anybody experience something similar?
Robert (view profile)
Terrific function. Very ease to use. Thanks a huge lot!
Conner Herndon (view profile)
Helen Smith (view profile)
Great function, however, is there a way to use if for displaying x error bars?
Jack Williams (view profile)
Great function!
I found that adding
if isdatetime(xP)
xP = datenum(xP);
end
at Line 144 meant that I could use datetime values as the x input.
orlyka (view profile)
Great function, thanks!
One question - the graphs I get are very heavy (6 MB and more when saved to PDF).
Any ideas on how to make them smaller?
Marie Mestdaghe (view profile)
My bad, I didn't clear my workspace and it influenced the x and y.
Rob Campbell (view profile)
Evan, I updated the submission to a GitHub page that I think already has the change you mention. I think it's narginchk, not nargoutchk, that you mean.
Rob Campbell (view profile)
I'll check out the 2016a issue.
Evan Smith (view profile)
Fantastic! Thanks!
Only, note that with MATLAB2016a you need to change line 53 to error(nargoutchk(3,5,nargin))
Matlab User (view profile)
Flexible suggestion for a legend:
for i = ...
h=shadedErrorBar(x_i, y_i, {@mean,@std});
h.mainLine.DisplayName = name_i;
end
legend(findobj(gca, '-regexp', 'DisplayName', '[^'']'));
Norman Shippee (view profile)
Is there a way to specify line and shading color outside of the base? The 'color' line property when specified as the [R G B] style is not accepted (or I am doing it wrong)
Rob Campbell (view profile)
Hello,
People have reported a few issues recently that I haven't seen until now. I'll get around to updating stuff soon.
Mark Benoit (view profile)
the left and bottom axes are a problem for me too. I just plotted a black line for the axes there. Not a great fix, but it works
Juan Carlos Moreno (view profile)
Hey Philip thanks for your reply.
Yes I try with set(gca,'YScale','log') and I can see the lien but the shading is gone.
anyway to fix this?
Philip Cheney (view profile)
Juan,
The simplest way to change the plot to semilogy is to add
set(gca,'YScale','log');
to the function. Unfortunately, it looks like the shading is lost in semilogy (though retained in semilogx), even though the error bounds are plotted. Is this an issue with patch?
Juan Carlos Moreno (view profile)
hi,
I am trying to use this function but need to do it in a semilog plot (semilogy). Can you tell me if there is anyway to do so?
thanks
E. Cheynet (view profile)
Hi Rob,
I was wondering if it is possible to add an option in your function allowing to flip the x and y axis ? Otherwise, that's a great function !
giacomo valle (view profile)
giacomo valle (view profile)
THIS FUNCTION DOESNT WORK:
Out of memory. The likely cause is an
infinite recursion within the program.
Gokhan Kirlik (view profile)
Marcin Konowalczyk (view profile)
Some suggested updates:
- use 'narginchk' instead of 'nargchk'
- use uistack(H.main_line,'top'); instead of deleting and re-plotting the line
My short version (with some removed functionality):
function H = shplot(x,y,e,varargin)
%% shplot(x,y,e,opt,varargin)
% Striped down version of shplot
% Based on a code by Rob Campbell
% Written by Marcin Konowalczyk
% Timmel Group @ Oxford University
x = x(:); y = y(:);
isheld = ishold; if ~isheld; cla; hold on; end;
H.main_line = plot(x,y,varargin{:});
eu = y + e; el = y - e;
col = 0.15*get(H.main_line,'color') + 0.85; ecol = 3*col-2;
set(gcf,'renderer','painters');
H.patch = patch([x ;flipud(x)],[el ;flipud(eu)],1,'facecolor',col,'edgecolor','none','facealpha',1);
H.edge_u = plot(x,eu,'-','color',ecol);
H.edge_l = plot(x,el,'-','color',ecol);
uistack(H.main_line,'top');
if ~isheld; hold off; end;
end
Adam (view profile)
Luis Felipe (view profile)
Nice!
Douglas (view profile)
The lines and the shaded region are not the same color for me, as in the example.
Yasamin Khorramzadeh (view profile)
Thanks for the helpful function. However, it seems like the shaded colors disappear when I go to logscale? Is there a way to fix this?
CDev (view profile)
Sorry I meant to say left and bottom axes lines rather than x axes.
CDev (view profile)
Thank you for this function! However when I set my shaded errors bars to be transparent the left and bottom x axes disappear. They reappear if I go set(gca, 'linewidth', 2) thereby making the axes thicker, but the right and top axes then are disproportionately think.
Has anyone figured out how to avoid the axes disappearing or bring them back once they've disappeared?
Matt (view profile)
Great, thanks!
Hugo Gonzalez Villasanti (view profile)
Alberto Jr (view profile)
Very new to Matlab and do not understand how to use this in my code? Can anyone explain step by step how to use it? I spent hours and I haven't succeeded in plotting one graph yet. Thank you
Diana (view profile)
YMusic: This is what I do to save the figure.
h=figure, shadedErrorBar(Ctrl(:,1), Ctrl(:,2),Std_Ctrl(:,2),'-k', 1)
hold on,
shadedErrorBar(T5(:,1), T5(:,2), Std_T5(:,2),'-y',1)
print(h,'-djpeg','-r600','FileName')
I hope it is useful for you.
YMusic (view profile)
Fantastic plots, BUT.... I can't get savefig or print to work... in your description you say to "plot2svg (FEX submission 7401) if you wish to get high quality printed files", but do you know somewhere that I could find an example of you doing this? I can't get it to work and it says that it can't save a struct. I tried changing it using num2cell and struct2table. I have ~200 shadederrorbars on one figure. Any guidance is appreciated. Thank you in advance!
korkam (view profile)
F A (view profile)
Rob Campbell (view profile)
Mich: the command returns the handles to the plot objects so to increase line thickness you can modify the plot object properties: http://www.matlab-cookbook.com/recipes/0050_Plotting/0010_Plot_Manipulation/029_whatIsAHandle.html
Sergei P. (view profile)
Mich (view profile)
I managed to add legend for my graph. I missed out one of Campbell's previous comment.
Here is it for those who might face the same pbl! :
A=shadedErrorBar(x,y,{@mean,@std},'-b',1);
hold on
B=shadedErrorBar(x,y+10,{@mean,@std},'-r',1);
>> A %This is what's in the handles
A =
mainLine: 174.0065
patch: 175.0050
edge: [176.0026 177.0026]
legend([A.mainLine,B.mainLine],'A','B')
Mich (view profile)
Thank you so much for this code!! Helped me alot! :D
Can i check how should i add legend and increase the thickness of the main line? I tried reading the comments to get some help but i dont quite uds.
Stefan Grandl (view profile)
Rob Campbell (view profile)
Sarah -- try the boundedline submission listed on the "This file inspired" bit in the File Information, above.
Sarah S. (view profile)
Thank you for this function. How can I do the same thing if the shaded variable is the x and not y? (plot of altitude vs concentration for example, where the concentration has a range of values to be calculated with the shadederrorbar function)
Thanks!
Kay (view profile)
Very nice, very helpful.
I have added ",'HandleVisibility','off'" to the patch and the edge lines (code lines 146,150 and 151). In that way, one can use "legend" in a created plot, and only the main lines are listed.
Nicolas (view profile)
Michael (view profile)
Super awesome. Easy to use. Beautiful graphics.
Andreas Trier Poulsen (view profile)
Tihana (view profile)
very helpful! Thank you.
Irene Seco (view profile)
Alex (view profile)
Rob Campbell (view profile)
Constantinos,
Etsi?
y=randn(30,80)+10; x=1:size(y,2);
shadedErrorBar(log(x),y,{@median,@std},{'r-o','markerfacecolor','r'});
Constantinos Constantinos (view profile)
Thanks for this code, it is a very nice way to represent standard errors. I just have an issue I can't solve: In one of my graphs, I would like to use a logarithmic scale for the x axis.. as soon as i change it from linear to log on the figure manager (or if i use set(gca,'xscale','log')), the axis change to log but the graph remains the same.. I have the feeling the function is not compatible to log graphs. do you have any idea how i could solve this problem? i tried to replace all "plot()" in your code with "semilogx()" but it also did not work.
Thanks in advance
Rob Campbell (view profile)
Bernadette, set(gcf,'Renderer','Painters')
http://www.mathworks.com/help/matlab/ref/figure_props.html
Bernadette (view profile)
thanks for this nice code, unfortuntately i can't figure out this openGL thing.. you wrote that if it's saved as.eps then it will look like a raster, and that's what it does in my plot but I don't know how to change it?! can you help?
Stone (view profile)
Very powerful and convenient to use!
Slavi (view profile)
Now thank you very much, your second option works perfect. This is what I need. I wish I can give you 5 stars now :).
You should include this in the description. It's really helpful :).
Thank you one more time
Rob Campbell (view profile)
Slavi, the behavior you see is is typical of functions such as this because each line is made up of multiple plot elements and the legend command doesn't know anything about that. There are two ways for you to proceed. Firstly, use the "axes" command in order to make a set of legend axes, then build your own legend within these axes. It'll be a bit of pain, but you get the most flexibility this way. Secondly, you can tell the legend command which plot elements to use:
A=shadedErrorBar(x,y,{@mean,@std},'-b',1);
hold on
B=shadedErrorBar(x,y+10,{@mean,@std},'-r',1);
>> A %This is what's in the handles
A =
mainLine: 174.0065
patch: 175.0050
edge: [176.0026 177.0026]
legend([A.mainLine,B.mainLine],'A','B')
Slavi (view profile)
Very helpful :). But unfortunately I can't display only the Y-values in the legend. There are more additional data sets in the legend I can't delete.
Do you have any idea? :)
cpas1940 (view profile)
uncung fgv (view profile)
what is shade error bar? i try to execute it but cannot.
Wantong (view profile)
Diego (view profile)
Franklin (view profile)
German Gomez-Herrero (view profile)
Agatha lenartowicz (view profile)
Very helpful function. Thanks!
Estella Liu (view profile)
EMC_solver (view profile)