plt

An alternative to plot and plotyy optimized for data exploration

Description

--- plt version 02Dec11 ---
An alternative to MatLab's plot/plotyy for 2D applications.
Like plot, plt commands can be typed at the command prompt to
display your workspace arrays and can also be the core of
your .m graphical applications. Some plt's advantages are:

Optimized for data exploration
Improved zooming, panning, and auto-scaling controls
Workspace plotting (interactive selection of variables to plot)
Automatically generated legend with powerful trace selection controls
Up to 99 traces on a single axis (or more if a legend isn't needed)
Support for dual y-axes (left/right linkable)
Support for sub-plots, each with individual cursor readout.
Fast and easy cursor movement
Delta cursor provides easy difference readout
RMS, mean, y/x ratio & abs(x+yi) readouts
Peak/Valley finder
Better looking grid lines with selectable color
User-entered annotations
User editing of trace properties and figure colors
Mouse and keyboard driven data modification
Support for metric prefixes
Quick linear/log toggling for x or y axes
Display expansion history
Flexible command-line interface
Includes advanced GUI components with sample GUI programs
Extensive on-screen help with 22 example script and function .m files
Congenial "customer" support & receptive to suggestions from users
Compatible with R12.1 or later

• MATLAB release: MATLAB 7.9 (2009b)
• Other requirements: Matlab version R12.1 or later

[Gil Hornung]

Very nice.
I especially like the fact that I can easily find out what are the values of some point.

[Claus Patz]

 

[Ilya Papiashvili]

The idea to enhance the plot is right, but exact type of the additional functionality as well as its implementation may vary. The tool may be more or less convinient for different kinds of tasks. However there is no possibility to adjust it or improove and develop farther, not to fix some bugs, since the .m file is not provided.

Conclusion - limited usability.

[Ben Michell]

I would agree that supplying the m-file would make it more usable for everyone.

Something I would particularly like to add/see added would be the ability to recognise structural arrays. E.G. if I have a structural array in the workspace typing 'plt' would give me the option of plotting any of the numeric varibles within the struct.

[Jiro Doke]

I think the features in this function are very impressive, and I especially like the smooth panning and zooming functions. The purpose of this function should be a visual examination of data, as opposed to a replacement of PLOT. It may not be suitable for creating publication-ready plots, as the appearance is not easily modifiable. Great for data exploration, though.

[Ben Michell]

Still using and enjoying plt. Have recently started using it for an application where I have variable names that are the names of people in the form firstname_surname.

Basically I'd like the full name of the variable to appear as the TraceID. Would it be possible to have that added or get a copy of the m-file so I can modify it myself?

Ben

[John Baldwinson]

I expect that a lot of us would need to modify the code for our own publication purposes. Hence usefullness greatly reduced by lack of source code.
If you intend to sell the code then get on with it.

[Paul Mennen]

> Hence usefullness greatly reduced by lack
> of source code. If you intend to sell
> the code then get on with it.

No John, I don't plan on selling the code.
The source wont actually do you much good
anyway since the code is not well commented
and is difficult to modify. You might be
better off asking me to make the changes
you need. However I will make the source
code available on request (no charge).
~Paul

[Marc Passy]

No support for dates as X axis (that I can find). Makes this useless for timeseries plotting.

[Paul Mennen]

> From: Marc Passy([e-mail])
> No support for dates as X axis (that I
> can find). Makes this useless for
> timeseries plotting.
Marc, you may want to try the latest update. Actually, I developed plt for an application involving time series exclusively. From my data exploration focus, date ticks were cumbersome for precise rendition of a time series, so dealt with dates and times outside of plt. However, recently I made this part of the plt package. Try running demo\pltn.m - note how the date of the cursored point is shown (resolution to 0.001 second!). Also type in a date or time and the cursor moves to that spot. I no longer use plot() for data exploration, but I would still use it for publishing a chart that could be best interpreted with date ticks. (I've dreamed that plt could totally replace plot, there probably will always be room for both).

[Snorri Ingvarsson]

Great program! I use it frequently for quickly viewing and comparing plots of variables in my workspace. It lets you pick variables with a mouseclick and plot by hitting a button. Then you can remove/add these from the view with a click and you can zoom/pan with the mouse (no going to a toolbar to pick "zoom" and then "pan". Handy when e.g. comparing peak positions in plots.

Thanks alot!

[Adam Groves]

When I downloaded and un-zipped the archive my AntiVirus popped up saying plt_.dll had a virus in it and promptly deleted it.

Is this an actual Virus or is it the stupid Virus software (Sophos) on my work computer?

[Adam Groves]

Please ignore my Previous Post it was my Antivirus flagging it up unnecessarily; there is no Virus present in this package.

Well, now I have tried it I don’t know where to start, I am a complete Newbie to Matlab and after seeing what PLT is capable of I don’t know where to start.

One thing is clear from the start, PLT is much more capable at plotting whatever type of data you have than the basic plot functions you get with Matlab from the off.

Now I just need to figure out how to use this great software to its full potential.

ps.

Thanks for all your help Paul

[Paul Mennen]

> Is this an actual Virus or is it the stupid Virus software
> (Sophos) on my work computer? Adam Groves

Adam - regarding your question about the Virus, I'm not sure the rating section is the best place for it. A direct email to me, or even the Matlab newsgroup would be more appropriate.
At any rate, I figured out why your anti virus software complained about plt_.dll. This dll is compressed using a method that Sophos is not aware of, so it automatically flags it as bad even though the dll does NOT contain a virus.
It sounds like you got over this particular problem, but never the less, I removed this dll from the most recent release because of the concern it has generated. The dll is not really needed (see the release notes) and I could always email it to anybody who requested it.
 

[Michael Hess]

Hi Paul,

This plot function is very good. You can scroll and zoom in easily. The only problem is that high data signals (e.g 60000 samples) are plotted incorrectly on the right scale.
This problem appears both when I use the plt(x,y) function and the plt function.
Is their a way to fix this problem or do I do something wrong?
Kind Regards,
Michael

[Paul Mennen]

Michael wrote: The only problem is that high data signals (e.g 60000 samples) are plotted incorrectly on the right scale.

Michael, I think the newsgroup or a private email would be a better place to discuss bugs as this section is mainly for ratings. But I'll try to answer your question anyway. I chose to plot the traces on the right hand axes using the "xor" line drawing mode which means that when it crosses another trace it inverts the overlapping pixels. This has some advantages when overlaying traces, although as you found it does have one disadvantage. When the trace overlaps itself (which it does many times in the 60,000 sample trace of your example), the trace can cancel itself out in places. It turns out that any method of displaying so many samples in so few pixels is misleading although you may argue that the xor method is not the best for this. I'm actually considering going back to normal mode for the right hand traces, although it will take me some time to verify that everything still works in a consistent manner. In the mean time I'll send you a work around which will allow you to change the plotting mode to the one you prefer.

[Michael Hess]

The plt function plots all my signals very well. I can change the colors and chose signals, find the peaks and dips and zoom in and out.
It is a very nice program. Thanks Paul for helping me with my questions.
Kind Regards, Michael

[John Lintern]

 

[Thierry Dalon]

Like John Baldwinson said: unfortunately it is pcode (no source code).
This is to my opinion contrary to FEX philosophy!
When error occurs you get stacked.
You can not modify/customize the code.
You don't learn anything from it.

[Thierry Dalon]

I correct: it was not obvious.
Source code is available at author's web site:
http://www.mennen.org/plt/plthome.htm

[Qes M]

Excellent graphing add-on when you want to get on and do other coding. Not difficult to use with variety of features and very helpful examples.

[R. J.]

It claims to have a virus...

[Paul Mennen]

I was wondering why the downloads had dropped off. Come on RJ, be fair. If you had bothered to read the other reviews you would have realized that there is no virus in this submission. Your incompetent anti-virus software will claim that any compressed DLL has a virus. The last time this happened I said I would remove the DLL since I didn't think anybody would still be using it. However I found a few users still were using it, so I put it back in ... and what do I get? Another thoughtless knee jerk reaction. Wouldn't it have been wiser to first contact the author?

[Husam Aldahiyat]

No virus in sight.

[Thierry Dalon]

Thank you Paul for the regular updates and specific user requests implementations :-) PLT is a great plot utility!

[Luis Martinez]

Very nice graphic tool. Easy to use and understand and very useful when dealing with graphics of different scales. Thanks Paul also for the advices in how to use 'plt', cheers.

[Mario Liverpool]

Very good job indeed Paul, ur a champion! thank u, now I have plenty of options to plot my graphs with!

[Nitika]

Thanks Paul, your tool is very convenient for quick look up of data. However, if you want 'journal' quality black & white plots, it requires a lot of work to make it look like you want it to. That way plotyy is easier to handle once you are able to get what you want on the right axes. Perhaps, you can look into getting a easy switch for 'regular' looking plots in Matlab standard format.

[Paul Mennen]

Nitka's comment: "Perhaps, you can look into getting a easy switch for 'regular' looking plots in Matlab standard format."

Hi Nitka. I'll bet you didn't bother reading the chm help file. Or if you ran the demoplt.m example you would have seen "gauss.m" which plots using the default Matlab colors. (Just include 'COLORdef','default' in the parameter list). I'm assuming that's what you mean by 'regular' looking, but I can assure you that plt gives you more flexibility in your color choices than the standard Matlab plot routine. (For example, in plot you can't set the grid color independently from the axis tick lables, which forces you into what many consider to be ugly overpowering grid lines). Try clicking on the "Help" tag in any of the example programs, then go to the 3rd section (command line arguments), and click on "Colors".

[Jeremy D]

Great addition to Matlab! This program is a huge improvement on plotting data as well as interactively explore it. The help files and demos are also very useful. It is worth learning how to use. If you use multiple y-axes plots or large amounts of data or you just want to explore data in graphical format quickly then plt is for you.

[Robert Barrie]

Has this been tested on R2009b?
At first I got an invalid property error on 'share' in demoplt at 14. I removed that parameter and then get
Error using ==> plt (Too many input arguments. pltn at 48)
when trying to run pltn(20)

it looks good but just thought I would ask?

[Paul Mennen]

> Robert Barrie wrote: "Has this been tested on R2009b?"
Thanks for the comment Robert. I found access to a system with R2009b, and indeed there was an issue. I fixed that, so if you download plt today it should work for you. It will likely work with R2010 as well, although since I don't have access to that, I'm hoping one of the plt users with that update will add a comment about that. ~Paul

[S]

big help. thanks for the files.

[Brett]

Absolutely amazing. I got frustrated trying to zoom in on certain parts of graphs I was trying to analyze, came to the user community and found this gem.

The only complaint I have is that when you are graphing one or two lines, the right click zoom and left click drag work smoothly. However, when you use 3 or more lines with thousands of data points the program slows down considerably (upwards of 4-5 second lag time before the graph responds.) This could reflect my computer's performance, obviously.

Aside from that you'd be a fool not to download this if you deal with graphing in matlab.

[Paul Mennen]

Brett - thanks very much for the high rating and the positive review. Regarding the slow update rate, you would have the same problem with plot, plotyy or other matlab plotting routines since like plt they all rely on the same underlying matlab objects (figure, axis, line, etc). That said, I'm curious how you get a 4 second lag. You say "thousands of points". Perhaps you were really into the millions?. For example, try this simple script:

  > t = (0:250000)/250000;
  > y1 = exp(-2*t).*sin(20*t);
  > y2 = t .* cos(5*pi*(1-t).^3);
  > y3 = exp(-1.4*t).*sin(10*pi*t.^5);
  > plt(t,[y1; y2; y3; humps(t)/100]);

This plots 4 traces, each of which has a quarter of a million points. Quite a stress for the plotting routines yet the update rate on my 6 year old computer (even modest in its day) is around .5 seconds - noticeable but still quite usable. So I suspect there is something else causing your slowdown. If you want to send me your code and/or data (at the email address inside the plt help file) I might have a suggestion for you. For very large data sets decimating the data before plotting may be necessary if smooth zooming/panning is desired.
~Paul

[Brett]

I believe that my slowdown was not related at all to the plt program. Ignore any complaints I had in my previous post.

[Nick Sinclair]

Generally spectacular for exploring data. Sadly, I want its functionality for plots I can publish, so I won't likely use it too much.. but for just displaying data in a very flexible way, it is fantastic.

It doesn't seem like it would be too hard to add a publish feature that would take away the menu items from the figure window and change the bg color to white, etc. Perhaps not, though. Great job, anyway!

[Paul Mennen]

> Nick wrote: It doesn't seem like it would be too hard to
> add a publish feature that would take away the menu items ...

Looks like you have noticed my bias towards data exploration, yet I haven't ignored the publishing angles entirely. For example, after you get everything looking the way you would like to publish the figure, right click on the y-axis label and the data exploration items (i.e. menu box, cursor controls, cursor readouts) disappear so as not to clutter up your copy.

Also you can include the 'colorDef' command to specify the colors you would like to use. The simplest choice is 'colorDef','Default' or 'colorDef',0 which selects the usual Mathworks' color scheme which isn't bad for publication use. If you start plt with its default dark background colors the screen colors will get inverted when you do a hardcopy (since printing with a black background generally produces miserable results).

For years I've been trying to get plt users to run the demo program (demoplt.m) at least once, yet amazingly enough when I ask new users they nearly always tell me they haven't run it.

That's unfortunate since one of the example plots would usually show the user that the feature they were pining for is available, and shows them the example code that does it. In your case, the appropriate example is called "pub" which shows two plots optimized for publication. These examples are also described in the .chm help file, although of course nobody reads manuals any more (even me) ... and I'm used to that by now :)

Anyway Nick, thanks for your review as well as the 5 stars!

~Paul

[Nick Sinclair]

busted! i didn't run the demo. got blinded by all the features I guess. I'll have another look at it. Thanks for the prompt help!

[Marc Cote]

Is it possible to link multiple traces to one color? If so can you point me an example that I could use? I'm interested to plot clusters and use the colors to hide/show them.

EDIT: I found out that pltquiv.m seems to be doing what I needed.

Great tool by the way.

[Paul Mennen]

> Marc wrote: Is it possible to link multiple traces to one color?

Hi Marc ... Thanks for the great rating.
  And yes you can set the trace colors quite easily. Note that if you type "plt(rand(6,30))" the 6 traces will appear in a different color for each trace, but if you type "plt(rand(6,30),'TraceC',[0 1 0]) then all 6 traces will be green. or use 'TraceC',[0 1 0; 1 0 0] and the odd numbered traces will be green and the evens will be red. Hope that helps :)
  ~Paul

[Brian Esler]

Awesome stuff!
I've just started experimenting with this type of sliderbar interface, and it's incredibly useful. You're functions were extremely instructive.

[Warwick]

This is great. So many ways it is better than plotyy. Is there an easy way I can set a default to publication figure so I don't have to type in "Colordef, 0, 'traceID, '', Options, '-All NoCursor' each time?
If not, it's a small price to pay for the time I have saved compared to the "old" way.

[Paul Mennen]

> Warwick wrote: Is there an easy way I can set a default to a publication figure?

Certainly Warwick. Create a new file in the plt folder called pltpub.m which contains this:

function out = pltpub(varargin)
  out = plt(varargin{:},...
  'ColorDef',0,...
  'TraceID',0,...
  'Options','-AN');

Then for example you can type something like "pltpub(x,[y1;y2;y3])" and those 3 traces will be plotted using parameters appropriate for generating publication hardcopy.

Thanks for your rating and comments Warwick.

[Warwick]

Many thanks, Paul.