Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.graphics.apps.gnuplot > #990 > unrolled thread

missing date in input file

Back to article view | Back to comp.graphics.apps.gnuplot

Contents

  missing date in input file Frank Ursel <sexyspamcatcher@trash-mail.com> - 2012-03-21 08:53 +0000
    Re: missing date in input file Hans-Bernhard Bröker <HBBroeker@t-online.de> - 2012-03-21 12:13 +0100
      Re: missing date in input file Frank Ursel <sexyspamcatcher@trash-mail.com> - 2012-03-21 14:04 +0000
        Re: missing date in input file Hans-Bernhard Bröker <HBBroeker@t-online.de> - 2012-03-21 16:18 +0100
    Re: missing date in input file bryan <bryanlepore@gmail.com> - 2012-03-21 07:01 -0700

#990 — missing date in input file

Hi,

I have a generated file with dates and values:

2012-01-01 10
2012-01-02 5
2012-01-05 15

As you can see, the dates are not continuous but should be printed 
continuous - and gnuplot does the right thing here.

Is there a way to tell gnuplot to interpolate/interprete the missing 
values for by gnuplot inserted dates as e.g. zero? Because I want to 
print the data like

plot "data" using 1:2 with lines

and this connectects 5 and 15 in my example but I would like to see a gap 
(as it is "with impulses") because the two missing days in x-axis are zero 
- and because of this are missing.

I know about missing data I could put a ? in the dataset but it's hard 
for me to figure out the missing dates and include the right ones, so I 
would like to tell gnuplot to have data values of non existing values 
interpret as zero or not to connect datapoints if there are by gnuplot 
inserted values between them.

Thanks,
Frank

[toc] | [next] | [standalone]

#991

On 21.03.2012 09:53, Frank Ursel wrote:

> Is there a way to tell gnuplot to interpolate/interprete the missing
> values for by gnuplot inserted dates as e.g. zero?

No, because there are no missing values in that dataset.  gnuplot cannot 
know the difference between a data point just not existing and one being 
omitted on purpose.  If you want such a difference to be made, you have 
to tell gnuplot about it explicitly --- by putting missing points into 
the file, marked appropriately.  Or put blank lines in there.  Those 
will also break 'with lines'.

> I know about missing data I could put a ? in the dataset but it's hard
> for me to figure out the missing dates and include the right ones,

If it's hard even for you, how do you expect gnuplot to manage it?

[toc] | [prev] | [next] | [standalone]

#993

Hans-Bernhard Bröker schrieb:
> On 21.03.2012 09:53, Frank Ursel wrote:
>> Is there a way to tell gnuplot to interpolate/interprete the missing
>> values for by gnuplot inserted dates as e.g. zero?
> 
> No, because there are no missing values in that dataset.  gnuplot cannot
> know the difference between a data point just not existing and one being
> omitted on purpose.

But it could know it - it knows how to draw the x-axis with dates, 
without gaps, even if some dates are missing in the data set.

If you want such a difference to be made, you have
> to tell gnuplot about it explicitly --- by putting missing points into
> the file, marked appropriately.  Or put blank lines in there.  Those
> will also break 'with lines'.

Ok, I will try to manage it this way.

Thanks,
Frank

[toc] | [prev] | [next] | [standalone]

#994

On 21.03.2012 15:04, Frank Ursel wrote:

> But it could know it - it knows how to draw the x-axis with dates,
> without gaps, even if some dates are missing in the data set.

No it can't, because there's no way it can know _why_ those data points 
are not present in your dataset.

For something to be missing, there has to be some expectation for it to 
be there, which then fails to be fulfilled.  gnuplot doesn't have any 
such expectations, thus no concept of something missing.

[toc] | [prev] | [next] | [standalone]

#992

On Wednesday, March 21, 2012 4:53:51 AM UTC-4, Frank Ursel wrote:
> 2012-01-01 10
> 2012-01-02 5
> 2012-01-05 15
> 
> I would like to see a gap (as it is "with impulses") because the two missing >days in x-axis are zero 

have you considered using a histogram? or layering different plot types?

MAYBE a ternary operator will help you skip the zeros - not sure how though. try `help ternary`.

HTH

-Bryan

[toc] | [prev] | [standalone]

Back to top | Article view | comp.graphics.apps.gnuplot

csiph-web