Reference line from an equation

[Gary Rosin]

I frequently generate scatterplots using the Chart-builder in the GUI. To
add a *non-linear* fit-line. The GUI for  the Chart-builder does not have an
option for this, so I use the Chart Editor, and manually "add a reference
line from an equation."

I'm trying to generate syntax to put in, at the time the graph is generated,
a line using the equation I need. The Chart Editor doesn't allow pasting
syntax. The Syntax Reference doesn't mention whether this is possible, or
how to do it. I've searched the archives, and haven't found any mention of
syntax for equation lines.

I haven't played around with either GPL or the Data Visualizer, which seems
even more opaque.

Does anyone know how to do this?

Gary

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Andy W]

Generate a synonymous scatterplot through the chart builder and paste the GPL code, it will look something like below;

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=VAR00001 VAR00002 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: VAR00001=col(source(s), name("VAR00001"), unit.category())
  DATA: VAR00002=col(source(s), name("VAR00002"))
  GUIDE: axis(dim(1), label("VAR00001"))
  GUIDE: axis(dim(2), label("VAR00002"))
  SCALE: linear(dim(2), include(0))
  ELEMENT: point(position(VAR00001*VAR00002))
END GPL.

Then, assuming a linear OLS fit line, add in the line after the point ELEMENT;

  ELEMENT: line(position(smooth.linear(VAR00001*VAR00002)))

See the GPL reference guide for what other types of lines are available (if you had looked they have an example of doing this exact thing). I can understand not reading it front to back - but it is quite painless to look through the examples to at least see what is possible!

[Gary Rosin]

Thanks, but I need to insert a non-linear fit line, so I need to specify the
equation.

Gary

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Andy W
Sent: Friday, April 26, 2013 3:44 PM
To: [hidden email]
Subject: Re: Reference line from an equation

Generate a synonymous scatterplot through the chart builder and paste the
GPL
code, it will look something like below;

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=VAR00001 VAR00002
MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: VAR00001=col(source(s), name("VAR00001"), unit.category())
  DATA: VAR00002=col(source(s), name("VAR00002"))
  GUIDE: axis(dim(1), label("VAR00001"))
  GUIDE: axis(dim(2), label("VAR00002"))
  SCALE: linear(dim(2), include(0))
  ELEMENT: point(position(VAR00001*VAR00002))
END GPL.

Then, assuming a linear OLS fit line, add in the line after the point
ELEMENT;

  ELEMENT: line(position(smooth.linear(VAR00001*VAR00002)))

See the GPL reference guide for what other types of lines are available (if
you had looked they have an example of doing this exact thing). I can
understand not reading it front to back - but it is quite painless to look
through the examples to at least see what is possible!



-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Reference-line-from-an-equatio
n-tp5719770p5719771.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[David Marso]

Understand from the start that I am a COMPLETE NOOB at GGRAPH and there is
likely a better/easier way .
(see Andrew's approach -He is one of the GPL wizards in these parts).
RTFM and Apply Andrew's insights and maybe my simulation idea?
*<Evil HACK> -.

MATRIX.
SAVE (UNIFORM(10000,1)) / OUTFILE * / VAR X.
END MATRIX.
COMPUTE Y=.7 + .3 * X + .5 * X**2 + LN(X) + NORMAL(1).
COMPUTE FIT=.7 + .3 * X + .5 * X**2 + LN(X).

GGRAPH
 /GRAPHDATASET NAME="graphdataset" VARIABLES=X Y FIT MISSING=LISTWISE REPORTMISSING=NO
 /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
 SOURCE: s=userSource(id("graphdataset"))
 DATA: X=col(source(s), name("X"))
 DATA: Y=col(source(s), name("Y"))
 DATA: FIT=col(source(s), name("FIT"))
 TRANS: X_Y=eval("X - Y")
 TRANS: X_FIT=eval("X - FIT")
 ELEMENT: point(position(X*Y), color.exterior(X_Y))
 ELEMENT: point(position(X*FIT), color.exterior(X_FIT))
END GPL.
*</EVIL HACK >.

Gary Rosin wrote

Thanks, but I need to insert a non-linear fit line, so I need to specify the
equation.

Gary

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Andy W
Sent: Friday, April 26, 2013 3:44 PM
To: [hidden email]
Subject: Re: Reference line from an equation

Generate a synonymous scatterplot through the chart builder and paste the
GPL
code, it will look something like below;

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=VAR00001 VAR00002
MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: VAR00001=col(source(s), name("VAR00001"), unit.category())
  DATA: VAR00002=col(source(s), name("VAR00002"))
  GUIDE: axis(dim(1), label("VAR00001"))
  GUIDE: axis(dim(2), label("VAR00002"))
  SCALE: linear(dim(2), include(0))
  ELEMENT: point(position(VAR00001*VAR00002))
END GPL.

Then, assuming a linear OLS fit line, add in the line after the point
ELEMENT;

  ELEMENT: line(position(smooth.linear(VAR00001*VAR00002)))

See the GPL reference guide for what other types of lines are available (if
you had looked they have an example of doing this exact thing). I can
understand not reading it front to back - but it is quite painless to look
through the examples to at least see what is possible!



-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Reference-line-from-an-equatio
n-tp5719770p5719771.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Andy W]

Yep the logic is basically spelled out by David. The easiest way is to save the predicted values, and put them in a second ELEMENT call (if the data is more sparse you would likely want the fitted values to be a line as opposed to points.) Or you can make the data via TRANS statements in inline GPL. The key is you have to edit in your own second ELEMENT statement.

You can see a few more examples here, http://andrewpwheeler.wordpress.com/2013/04/03/some-notes-on-single-line-charts-in-spss/ (error intervals, background guidelines etc.)

[Gary Rosin]

Following Andy W's  earlier gentle suggestion to RTM, I finally
found the SPSS GPL Reference Guide. That has an example of the
GPL language that you need to add to the GGRAPH command.

It's even logical, once you buy a vowel--a graphic vowel. This
is my first foray into GPL.

First, you define a new variable using a DATA statement
and the iter function:

   DATA: x = iter(<from>, <to>, <step>)

Second, you use the TRANS statement and the eval function to
generate another variable:

   TRANS: y = eval(<expression>)

Last, you add the new graphic element:

   ELEMENT: line(position(x*y))

To add an exponential equation for x from 0 to 100, you'd add
to your GPL block:

   DATA: x = iter(0, 100, 0.01)
   TRANS: y = eval(exp(x))
   ELEMENT: line(position(x*y))

The nice thing about this, is that you don't change your working
dataset. The x and the y are just working variables.

So far as I can tell, there is no generic smooth function that
works with any equation. To simulate smoothness, you have to put
in smaller steps, which takes up more memory, and slows down
execution.

One downside is that you can't edit the equation in the Chart
Editor.

It would be nice to have an extension command that would both
drive the GPL addition, and make the line interactive, and
editable in the Chart Editor.

Gary


----- Original Message -----
From: "Andy W" <[hidden email]>
To: [hidden email]
Sent: Friday, April 26, 2013 7:27:28 PM
Subject: Re: Reference line from an equation

Yep the logic is basically spelled out by David. The easiest way is to save
the predicted values, and put them in a second ELEMENT call (if the data is
more sparse you would likely want the fitted values to be a line as opposed
to points.) Or you can make the data via TRANS statements in inline GPL. The
key is you have to edit in your own second ELEMENT statement.

You can see a few more examples here,
http://andrewpwheeler.wordpress.com/2013/04/03/some-notes-on-single-line-charts-in-spss/
(error intervals, background guidelines etc.)

<http://spssx-discussion.1045642.n5.nabble.com/file/n5719776/LineCharts1.png>







-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Reference-line-from-an-equation-tp5719770p5719776.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Jon K Peck]

What's wrong with CURVEFIT for this case if you just want an exponential fit?  Or the STATS REGRESS PLOT for a variety of polynomial or loess filts?  Both produce fully editable charts?


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Gary Rosin <[hidden email]>
To:        [hidden email],
Date:        04/26/2013 08:47 PM
Subject:        Re: [SPSSX-L] Reference line from an equation
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Following Andy W's  earlier gentle suggestion to RTM, I finally
found the SPSS GPL Reference Guide. That has an example of the
GPL language that you need to add to the GGRAPH command.

It's even logical, once you buy a vowel--a graphic vowel. This
is my first foray into GPL.

First, you define a new variable using a DATA statement
and the iter function:

  DATA: x = iter(<from>, <to>, <step>)

Second, you use the TRANS statement and the eval function to
generate another variable:

  TRANS: y = eval(<expression>)

Last, you add the new graphic element:

  ELEMENT: line(position(x*y))

To add an exponential equation for x from 0 to 100, you'd add
to your GPL block:

  DATA: x = iter(0, 100, 0.01)
  TRANS: y = eval(exp(x))
  ELEMENT: line(position(x*y))

The nice thing about this, is that you don't change your working
dataset. The x and the y are just working variables.

So far as I can tell, there is no generic smooth function that
works with any equation. To simulate smoothness, you have to put
in smaller steps, which takes up more memory, and slows down
execution.

One downside is that you can't edit the equation in the Chart
Editor.

It would be nice to have an extension command that would both
drive the GPL addition, and make the line interactive, and
editable in the Chart Editor.

Gary


----- Original Message -----
From: "Andy W" <[hidden email]>
To: [hidden email]
Sent: Friday, April 26, 2013 7:27:28 PM
Subject: Re: Reference line from an equation

Yep the logic is basically spelled out by David. The easiest way is to save
the predicted values, and put them in a second ELEMENT call (if the data is
more sparse you would likely want the fitted values to be a line as opposed
to points.) Or you can make the data via TRANS statements in inline GPL. The
key is you have to edit in your own second ELEMENT statement.

You can see a few more examples here,
http://andrewpwheeler.wordpress.com/2013/04/03/some-notes-on-single-line-charts-in-spss/
(error intervals, background guidelines etc.)

<http://spssx-discussion.1045642.n5.nabble.com/file/n5719776/LineCharts1.png>







-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Reference-line-from-an-equation-tp5719770p5719776.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD