gnuplot-beta

errorlines vs errorbars

[Allin C. <cot...@wf...>]

Is there any real difference between "errorbars" and "errorlines"?  
>From the help for errorbars it sounds as if there should be a 
difference, but I can't see any in the output.

One feature that would be nice to have -- and that, for a moment, 
I though "errorlines" might provide -- is the ability to plot a 
primary series plus y-axis error high and low using just lines 
(and not vertical bars) for all 3 components.  (This is the most 
common way of graphing forecasts with error bands in economics.)

Of course one can achieve this effect manually by providing y, 
high and low as 3 distinct arguments to 'plot' and forcing the 
high and low series to use the same line style/type/color.  But it 
would be a lot more convenient to be able to provide the same 
basic input and select the style via errorbars vs 
errorsomethingelse.

-- 
Allin Cottrell
Department of Economics
Wake Forest University, NC

Re: errorlines vs errorbars

[Ethan M. <merritt@u.washington.edu>]

On Tuesday 18 March 2008 07:08, Allin Cottrell wrote:
> Is there any real difference between "errorbars" and "errorlines"?  
> >From the help for errorbars it sounds as if there should be a 
> difference, but I can't see any in the output.
> 
> One feature that would be nice to have -- and that, for a moment, 
> I though "errorlines" might provide -- is the ability to plot a 
> primary series plus y-axis error high and low using just lines 
> (and not vertical bars) for all 3 components.  (This is the most 
> common way of graphing forecasts with error bands in economics.)

This sounds similar to the "between" option of "filledcurves".

> Of course one can achieve this effect manually by providing y, 
> high and low as 3 distinct arguments to 'plot' and forcing the 
> high and low series to use the same line style/type/color.  

In current cvs:

   plot for [i=2:4] 'data' using 1:(column(i)) with lines lc rgb "blue"

where 'data' contains    x ymin y ymax


> But it  
> would be a lot more convenient to be able to provide the same 
> basic input and select the style via errorbars vs 
> errorsomethingelse.
 
I think for quite a while now the philosophy has been that it is
sufficient to provide components from which you can build up a
complicated plot.  For example there is really no advantage of having
a style "linespoints" when you could make the same plot using
"lines" and "points" separately.  

-- 
Ethan A Merritt 

Re: errorlines vs errorbars

[Hans-Bernhard B. <HBB...@t-...>]

Allin Cottrell wrote:
> Is there any real difference between "errorbars" and "errorlines"?  

Yes.  "errorlines" relates to "errorbars" the same way "linespoints" 
does to "points".

>>From the help for errorbars it sounds as if there should be a 
> difference, but I can't see any in the output.

The output of... what?  Here's an example:

	plot '-' u 0:0:(1.0) w errorlines, \
              '-' u 0:($0+2):(1.0) w errorbars
	  0
           1
           2
           3
         E
	  0
           1
           2
           3
         E

> One feature that would be nice to have -- and that, for a moment, 
> I though "errorlines" might provide -- is the ability to plot a 
> primary series plus y-axis error high and low using just lines 
> (and not vertical bars) for all 3 components.  (This is the most 
> common way of graphing forecasts with error bands in economics.)

No.  If that were what it did, it would be called "errorband" or 
"corridor" ;-)

Re: errorlines vs errorbars

[Allin C. <cot...@wf...>]

On Thu, 20 Mar 2008, Hans-Bernhard Bröker wrote:

> Allin Cottrell wrote:
> > Is there any real difference between "errorbars" and "errorlines"?  
> 
> Yes.  "errorlines" relates to "errorbars" the same way 
> "linespoints" does to "points".
> 
> > From the help for errorbars it sounds as if there should be a 
> > difference, but I can't see any in the output.
> 
> The output of... what?  Here's an example...

Thanks, that clarifies things.  I now realize that I wasn't seeing 
a difference because in the plots I was looking at the point 
forecast was plotted as a line in its own right, hence supplying 
the "missing element" in the "errorbars" plot style for the 
standard errors.

> > One feature that would be nice to have -- and that, for a 
> > moment, I thought "errorlines" might provide -- is the ability 
> > to plot a primary series plus y-axis error high and low using 
> > just lines (and not vertical bars) for all 3 components.  
> > (This is the most common way of graphing forecasts with error 
> > bands in economics.)
> 
> No.  If that were what it did, it would be called "errorband" or 
> "corridor" ;-)

OK.  Then "errorband" would be nice to have, though not essential.

Allin Cottrell