3d plot of means

[Mario Giesel]

Hello, SPSS friends,

 

I'm trying to find a solution for this one - maybe somebody can help:

I've got means for companies on three dimensions.

I want to plot the companies as centroids within a 3d diagram with axes F01, F02, and F03.

Companies should be represented by points which are connected to F01-F03 plane by a line - i.e. "balancing on a stilt".

Any ideas with SPSS or R are welcome.

 

Here's the data:

 

DATA LIST FREE (" ")/company F01 F02 F03.

BEGIN DATA

1 -0,10843674 0,203262369 0,01296737

2 -0,105096601 -0,025509327 -0,091137073

3 0,222312714 0,057440887 -0,117708879

4 0,667471767 -0,137292544 -0,252940788

5 -0,122125046 -0,44443488 0,148065611

6 0,461112382 -0,216526175 0,0626079

7 0,060317046 -0,357790726 -0,060617492

END DATA.

EXE.

VAR LAB company "Company".

VAL LAB company 1 "A" 2 "B" 3 "C" 4 "D" 5 "E" 6 "F" 7 "G".

VAR LAB F01 "Property 1".

VAR LAB F02 "Property 2".

VAR LAB F03 "Property 3".

 

Thanks for any help,

Mario

[Art Kendall]

Unfortunately SPSS does not have a built in profile graph to display the results of clustering, repeated measures, etc.  Parallel coordinate plots are almost the same except they have different scales for each axis.

Below there are two sets of syntax the first does a profile chart it is a some syntax ViAnn helped me with.  It does a profile with 14 dimensions.  If this is what you are looking for you can just remove the extra 11 dimensions.  The scales on the two sides are crowded.

The second I just used chart builder to create a 3-D dot chart.

One you run a graph you can do all kind of editing with the chart editor in the output window.

Art Kendall
Social Research Consultants

data list list/ profile (f2) dim1 to dim14 (14f5.2).
begin data.
1    .00    .00    .00    .02    .04    .23    .72    .52    .01    .00    .00    .25    .26    .19
2    .04    .03    .00    .03    .02    .14    .46    .35    .04    .00    1.00    .17    .26    .13
3    .01    .00    .00    .02    .02    .12    .00    .33    .03    .01    .12    .10    1.00    .55
4    .00    .01    .04    .03    .02    .40    .89    .00    .62    .52    .57    .25    .12    .12
5    .00    .00    .03    .07    .03    .43    1.00    1.00    .46    .40    .66    .11    .05    .04
6    .07    .00    .02    .09    .04    .43    .89    1.00    .52    .48    1.00    .98    .00    .04
7    .00    .00    .00    .00    .00    .00    .00    .00    .00    .01    .00    .00    .00    .00
8    .46    .38    .36    .53    .27    .64    .81    .76    .62    .59    .71    .59    .58    .59
end data.

execute.
 
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=profile dim1 to dim14 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
   TEMPLATE=[
    "C:\Users\Vi\Desktop\test.sgt"].
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: dim1=col(source(s), name("dim1"))
  DATA: dim2=col(source(s), name("dim2"))
  DATA: dim3=col(source(s), name("dim3"))
  DATA: dim4=col(source(s), name("dim4"))
  DATA: dim5=col(source(s), name("dim5"))
  DATA: dim6=col(source(s), name("dim6"))
  DATA: dim7=col(source(s), name("dim7"))
  DATA: dim8=col(source(s), name("dim8"))
  DATA: dim9=col(source(s), name("dim9"))
  DATA: dim10=col(source(s), name("dim10"))
  DATA: dim11=col(source(s), name("dim11"))
  DATA: dim12=col(source(s), name("dim12"))
  DATA: dim13=col(source(s), name("dim13"))
  DATA: dim14=col(source(s), name("dim14"))
  DATA: profile=col(source(s), name("profile"), unit.category())
  TRANS: caseid = index()
  COORD: parallel()
  SCALE: linear(dim(1), min(-.1), max(1.1))
  SCALE: linear(dim(2), min(-.1), max(1.1))
  SCALE: linear(dim(3), min(-.1), max(1.1))
  SCALE: linear(dim(4), min(-.1), max(1.1))
  SCALE: linear(dim(5), min(-.1), max(1.1))
  SCALE: linear(dim(6), min(-.1), max(1.1))
  SCALE: linear(dim(7), min(-.1), max(1.1))
  SCALE: linear(dim(8), min(-.1), max(1.1))
  SCALE: linear(dim(9), min(-.1), max(1.1))
  SCALE: linear(dim(10), min(-.1), max(1.1))
  SCALE: linear(dim(11), min(-.1), max(1.1))
  SCALE: linear(dim(12), min(-.1), max(1.1))
  SCALE: linear(dim(13), min(-.1), max(1.1))
  SCALE: linear(dim(14), min(-.1), max(1.1))
  GUIDE: legend(aesthetic(aesthetic.color), label("profile"))
  GUIDE: axis(dim(2), ticks(null()))
  GUIDE: axis(dim(3), ticks(null()))
  GUIDE: axis(dim(4), ticks(null()))
  GUIDE: axis(dim(5), ticks(null()))
  GUIDE: axis(dim(6), ticks(null()))
  GUIDE: axis(dim(7), ticks(null()))
  GUIDE: axis(dim(8), ticks(null()))
  GUIDE: axis(dim(9), ticks(null()))
  GUIDE: axis(dim(10), ticks(null()))
  GUIDE: axis(dim(11), ticks(null()))
  GUIDE: axis(dim(12), ticks(null()))
  GUIDE: axis(dim(13), ticks(null()))
  ELEMENT: line(position(dim1*dim2*dim3*dim4*dim5*dim6*dim7*dim7*dim9*dim10*dim11*dim12*dim13*dim14),
  split(caseid), color(profile),
  transparency(transparency.".5"),size(size."5px"))
END GPL.

------------------------------
DATA LIST FREE (" ")/company F01 F02 F03.
BEGIN DATA
1 -0.10843674 0.203262369 0.01296737
2 -0.105096601 -0.025509327 -0.091137073
3 0.222312714 0.057440887 -0.117708879
4 0.667471767 -0.137292544 -0.252940788
5 -0.122125046 -0.44443488 0.148065611
6 0.461112382 -0.216526175 0.0626079
7 0.060317046 -0.357790726 -0.060617492
END DATA.
EXE.
VAR LAB company "Company".
VAL LAB company 1 "A" 2 "B" 3 "C" 4 "D" 5 "E" 6 "F" 7 "G".
VAR LAB F01 "Property 1".
VAR LAB F02 "Property 2".
VAR LAB F03 "Property 3".

* Chart Builder.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=F01 F02 F03 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: F01=col(source(s), name("F01"))
  DATA: F02=col(source(s), name("F02"))
  DATA: F03=col(source(s), name("F03"))
  COORD: rect(dim(1,2,3))
  GUIDE: axis(dim(1), label("Property 3"))
  GUIDE: axis(dim(2), label("Property 1"))
  GUIDE: axis(dim(3), label("Property 2"))
  ELEMENT: point(position(F03*F01*F02))
END GPL.

On 8/19/2011 1:55 PM, Mario Giesel wrote:

Hello, SPSS friends,

 

I'm trying to find a solution for this one - maybe somebody can help:

I've got means for companies on three dimensions.

I want to plot the companies as centroids within a 3d diagram with axes F01, F02, and F03.

Companies should be represented by points which are connected to F01-F03 plane by a line - i.e. "balancing on a stilt".

Any ideas with SPSS or R are welcome.

 

Here's the data:

 

DATA LIST FREE (" ")/company F01 F02 F03.

BEGIN DATA

1 -0,10843674 0,203262369 0,01296737

2 -0,105096601 -0,025509327 -0,091137073

3 0,222312714 0,057440887 -0,117708879

4 0,667471767 -0,137292544 -0,252940788

5 -0,122125046 -0,44443488 0,148065611

6 0,461112382 -0,216526175 0,0626079

7 0,060317046 -0,357790726 -0,060617492

END DATA.

EXE.

VAR LAB company "Company".

VAL LAB company 1 "A" 2 "B" 3 "C" 4 "D" 5 "E" 6 "F" 7 "G".

VAR LAB F01 "Property 1".

VAR LAB F02 "Property 2".

VAR LAB F03 "Property 3".

 

Thanks for any help,

Mario

===================== 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

[Pablo Reinoso]

  

Estaré fuera de la oficina a partir del viernes 19 al 31 de Agosto, por temas de soporte técnico comunicarse con Eduardo Caro, [hidden email],  por temas de capacitación comunicarse con Carlos Padilla [hidden email], otros temas favor contactar a [hidden email]. fono oficina:+56.2.4153478

 

[Jon K Peck]

Parallel coordinates are now directly available via the GraphBoard templates.  You will see them listed in the Visualization type control on the Detailed tab.  GraphBoard generates the appropriate GGRAPH syntax and does not require a special template such as the one (not) included in Art's post.



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




From:        Art Kendall <[hidden email]>
To:        [hidden email]
Date:        08/20/2011 05:47 AM
Subject:        Re: [SPSSX-L] 3d plot of means
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Unfortunately SPSS does not have a built in profile graph to display the results of clustering, repeated measures, etc.  Parallel coordinate plots are almost the same except they have different scales for each axis.

Below there are two sets of syntax the first does a profile chart it is a some syntax ViAnn helped me with.  It does a profile with 14 dimensions.  If this is what you are looking for you can just remove the extra 11 dimensions.  The scales on the two sides are crowded.

The second I just used chart builder to create a 3-D dot chart.

One you run a graph you can do all kind of editing with the chart editor in the output window.

Art Kendall
Social Research Consultants

[ViAnn Beadle]

GPL will draw the points and anchor the lines to the 0 point but will not render a plane transection the Y axis.  It draws the floor of x*z at the minimum point of Y.

 

Here’s my version:

 

GGRAPH

  /GRAPHDATASET NAME="graphdataset" VARIABLES=F01 F02 F03 MISSING=LISTWISE REPORTMISSING=NO

  /GRAPHSPEC SOURCE=INLINE.

BEGIN GPL

  SOURCE: s=userSource(id("graphdataset"))

  DATA: F01=col(source(s), name("F01"))

  DATA: F02=col(source(s), name("F02"))

  DATA: F03=col(source(s), name("F03"))

  COORD: rect(dim(1,2,3))

  SCALE: linear(dim(2), origin(0))

 

  GUIDE: axis(dim(1), label("F03"))

  GUIDE: axis(dim(2), label("F01"))

  GUIDE: axis(dim(3), label("F02"), gridlines())

  ELEMENT: interval(position(F03*F01*F02))

  ELEMENT: point(position(F03*F01*F02), shape.interior(shape.square), size(size."5px"), color(color.black))

END GPL.

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mario Giesel
Sent: Friday, August 19, 2011 11:55 AM
To: [hidden email]
Subject: 3d plot of means

 

Hello, SPSS friends,

 

I'm trying to find a solution for this one - maybe somebody can help:

I've got means for companies on three dimensions.

I want to plot the companies as centroids within a 3d diagram with axes F01, F02, and F03.

Companies should be represented by points which are connected to F01-F03 plane by a line - i.e. "balancing on a stilt".

Any ideas with SPSS or R are welcome.

 

Here's the data:

 

DATA LIST FREE (" ")/company F01 F02 F03.

BEGIN DATA

1 -0,10843674 0,203262369 0,01296737

2 -0,105096601 -0,025509327 -0,091137073

3 0,222312714 0,057440887 -0,117708879

4 0,667471767 -0,137292544 -0,252940788

5 -0,122125046 -0,44443488 0,148065611

6 0,461112382 -0,216526175 0,0626079

7 0,060317046 -0,357790726 -0,060617492

END DATA.

EXE.

VAR LAB company "Company".

VAL LAB company 1 "A" 2 "B" 3 "C" 4 "D" 5 "E" 6 "F" 7 "G".

VAR LAB F01 "Property 1".

VAR LAB F02 "Property 2".

VAR LAB F03 "Property 3".

 

Thanks for any help,

Mario

[Art Kendall]

I forgot to mention that you can workaround to get a profile chart by including extra profiles and using parallel coordinates plot.
1 extra profile has all values at the max you wish for all scales, e.g., 3.5 for z-scores, 100 for percents and percentiles, n for n tiles. etc.
1 extra profile has all values at the min you wish for all scales, e.g., - 3.5 for z-scores, zero for percents.  and 1 for percentiles an other n tiles, etc.
perhaps some reference lines e.g., for z-scores (-3,-2,-1,0,1,2,3), for percents and percentiles (5, 10,25, 50,  75, 90, 95) etc.

Then you can use the chart editor to make the reference frame all one color/pattern.

Art Kendall
Social Research Consultants

On 8/20/2011 10:10 AM, Jon K Peck wrote:

Parallel coordinates are now directly available via the GraphBoard templates.  You will see them listed in the Visualization type control on the Detailed tab.  GraphBoard generates the appropriate GGRAPH syntax and does not require a special template such as the one (not) included in Art's post.



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




From:        Art Kendall [hidden email]
To:        [hidden email]
Date:        08/20/2011 05:47 AM
Subject:        Re: [SPSSX-L] 3d plot of means
Sent by:        "SPSSX(r) Discussion" [hidden email]

---

Unfortunately SPSS does not have a built in profile graph to display the results of clustering, repeated measures, etc.  Parallel coordinate plots are almost the same except they have different scales for each axis.

Below there are two sets of syntax the first does a profile chart it is a some syntax ViAnn helped me with.  It does a profile with 14 dimensions.  If this is what you are looking for you can just remove the extra 11 dimensions.  The scales on the two sides are crowded.

The second I just used chart builder to create a 3-D dot chart.

One you run a graph you can do all kind of editing with the chart editor in the output window.

Art Kendall
Social Research Consultants

===================== 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