How do I can extrapolate data with gnuplot?

mfduqued

unread,

Nov 1, 2010, 2:17:28 PM11/1/10

to

Hi community.

Thanks for your time and your suggestions in the past.

In this moment, I need extrapolate data from my file data.prn. I can
fit this data with the command fit but I need to extrapolate data
about the vertical axis.

How can I do this?

Thanks for your colaboration and your suggestion.

MFDD

Hans-Bernhard Bröker

unread,

Nov 1, 2010, 5:19:53 PM11/1/10

to

On 01.11.2010 19:17, mfduqued wrote:

> In this moment, I need extrapolate data from my file data.prn. I can
> fit this data with the command fit but I need to extrapolate data
> about the vertical axis.

That doesn't make any sense. Extrapolation happens _along_ an axis, not
about it. And regardless what kind of fit you did, the vertical axis
can't the one you extrapolate along.

Maybe you need to step back a bit and re-start your explanation with
what you actually have, and what you want to do.

mfduqued

unread,

Nov 1, 2010, 6:54:54 PM11/1/10

to

Thank you for your time.

My problem is:

I want to know what is the value of intercept with the vertical axis.
My curve is linear (y = m * x + b) in a range (0< x< 0.5) and
exponential in another range (x>0.5).

This problem is because my data fluctuate close enough to x = 0,
therefore I wish to make a linear regression to extrapolate the point
b.

Thanks for your help

Hans-Bernhard Bröker

unread,

Nov 2, 2010, 4:46:11 PM11/2/10

to

On 01.11.2010 23:54, mfduqued wrote:

> On Nov 1, 4:19 pm, Hans-Bernhard Br�ker<HBBroe...@t-online.de> wrote:

>> Maybe you need to step back a bit and re-start your explanation with

> My problem is:
>
> I want to know what is the value of intercept with the vertical axis.
> My curve is linear (y = m * x + b) in a range (0< x< 0.5) and
> exponential in another range (x>0.5).

You've already fitted that intercept: it's "b".

mfduqued

unread,

Nov 2, 2010, 10:39:26 PM11/2/10

to

Mr. Bernhard

thank you for your time.

My problem is that my regression's graph is to the points that are
plotted, I want to eliminate noise points and extrapolate the other
points to the interception

Thank you for your suggestion

Karl

unread,

Nov 3, 2010, 4:44:58 AM11/3/10

to

On 03.11.2010 03:39, mfduqued wrote:
> My problem is that my regression's graph is to the points that are
> plotted, I want to eliminate noise points and extrapolate the other
> points to the interception

If you扉e got datapoints that are obviously wrong, why don愒 you
just remove them from your datafile?

regards, karl

P.S. It悲 be nice to have your name too, instead of a scrambled
"MFDD". Netiquette, it愀 called. :-)

mfduqued

unread,

Nov 3, 2010, 11:46:15 AM11/3/10

to

On 3 nov, 03:44, Karl <mail....@gmx.net> wrote:
> On 03.11.2010 03:39, mfduqued wrote:
>
> > My problem is that my regression's graph is to the points that are
> > plotted, I want to eliminate noise points and extrapolate the other
> > points to the interception
>

> If you´ve got datapoints that are obviously wrong, why don´t you

> just remove them from your datafile?
>
> regards, karl
>

> P.S. It´d be nice to have your name too, instead of a scrambled
> "MFDD". Netiquette, it´s called. :-)

Thanks Karl,

My problem is that:

I remove these points' noise but when I fit my points to my points
minus points' noise I don´t know what is the intercept?. This is
because my fit begins at my first point and this point isn´t at the
axis vertical.

How can I know this point?

Thank for your help

Mauricio F. Duque D.

Hans-Bernhard Bröker

unread,

Nov 3, 2010, 1:19:15 PM11/3/10

to

On 03.11.2010 16:46, mfduqued wrote:
> On 3 nov, 03:44, Karl<mail....@gmx.net> wrote:
>> On 03.11.2010 03:39, mfduqued wrote:

>>> My problem is that my regression's graph is to the points that are
>>> plotted, I want to eliminate noise points and extrapolate the other
>>> points to the interception

>> If you“ve got datapoints that are obviously wrong, why don“t you
>> just remove them from your datafile?

> I remove these points' noise but when I fit my points to my points

Err... when you do what?

> minus points' noise

I wonder how it comes that with every answer you write, you're making
even less sense than you did before. Nobody (including) said anything
about subtracting any points' noise.

> How can I know this point?

Which part of "you know it, it's 'b'" did you fail to grasp?

Karl

unread,

Nov 3, 2010, 3:19:25 PM11/3/10

to

On 03.11.2010 16:46, mfduqued wrote:

> I remove these points' noise but when I fit my points to my points
> minus points' noise I don´t know what is the intercept?. This is
> because my fit begins at my first point and this point isn´t at the
> axis vertical.

how do you remove the noise? i have a lot of noise on my data, too,
and i´d very much like to get rid of it.

kts

unread,

Nov 3, 2010, 11:47:07 PM11/3/10

to

is this what you are trying to do?

fit [0:0.5] a*x+b 'your-data' via a,b
plot [0:0.5] a*x+b

if you want to know the value of b, examine the file 'fit.log' , it's
near the bottom.

does that help?

mfduqued

unread,

Nov 4, 2010, 9:45:55 AM11/4/10

to

Thank you for your response, your suggestion is very useful

Mauricio F. Duque D.

mfduqued

unread,

Nov 4, 2010, 9:55:02 AM11/4/10

to

Mr. Hans-Bernhard.

Thanks for your time and I am sorry for my English.

Your comment help me that I don´t remove these points.

Thanks for your help

Mauricio F. Duque D.

sfeam

unread,

Nov 4, 2010, 3:48:02 PM11/4/10

to

kts wrote:

It's simpler than that. Just type "print b" after fitting

mfduqued

unread,

Nov 4, 2010, 9:10:53 PM11/4/10

to

thank you, very useful, very much