Mixed plot: Stacked Area + two Lines of different scale

[Kirill Orlov]

Please help me to plot this, if it is possible. Here is the data.

     age   height   weight   sexd_p

      10      134     38.0        0
      10      132     32.0        0
      10      146     84.0        0
      10      154     39.0        0
      10      132     29.0        0
      10      140     30.0        0
      11      142     43.0        0
      12      168     58.0        0
      12      143     52.0        0
      13      170     61.0        0
      10      141     31.5        1
      11      151     48.0        1
      12      150     41.0        1
      12      144     51.0        1
      13      160     55.0        1
      13      166     68.0        1
      14      170     55.0        1
      14      175     70.0        1
      14      173     76.7        1
      14      169     89.8        1
      14      168     58.0        1
      12      155     52.5        2
      13      160     65.0        2
      14      183     59.0        2
      14      173     56.0        2
      14      159     47.0        2
      14      168     53.0        2
      15      179     65.0        2
      15      177     88.0        2
      16      165     54.0        2
      14      161     58.0        3
      14      168     61.0        3
      14      159     44.0        3
      14      168     57.0        3
      14      163     62.0        3
      14      164     50.0        3
      14      173     66.0        3
      15      170     66.4        3
      15      171     50.0        3
      15      177     57.0        3
      15      166     70.0        3
      15      177     54.0        3
      15      170     55.0        3
      15      152     49.0        3
      15      156     50.0        3
      16      171     65.0        3
      16      175     62.0        3
      16      188     72.0        3
      16      159     66.0        3
      16      169     60.0        3
      16      191     80.0        3
      16      161     58.0        4
      16      185     67.0        4

AGE is ordinal or scale
SEXD_P (stage of sex development in pubic hairing) is ordinal
HEIGHT and WEIGHT are scale.

I need to plot AGE as X axis and SEXD_P as stacked area (For each age, this is the percent distribution of the stages summing to 100%).
On the same graph, I want to add two lines, the dynamics of HEIGHT (mean for each age) and WEIGHT (same). Since height and weight are
different units and scales of measurement I feel like giving dual Y axes: Y1 for HEIGHT, Y2 for WEIGHT.

In other words, I want to combine graph:

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=age COUNT()[name="COUNT"] sexd_p MISSING=LISTWISE
    REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: age=col(source(s), name("age"), unit.category())
  DATA: COUNT=col(source(s), name("COUNT"))
  DATA: sexd_p=col(source(s), name("sexd_p"), unit.category())
  GUIDE: axis(dim(1), label("Возраст"))
  GUIDE: axis(dim(2), label("Percent"))
  GUIDE: legend(aesthetic(aesthetic.color.interior), label("P - лобковое оволосение"))
  SCALE: cat(dim(1), include("10", "11", "12", "13", "14", "15", "16"))
  SCALE: linear(dim(2), include(0))
  SCALE: cat(aesthetic(aesthetic.color.interior), include("0", "1", "2", "3", "4"))
  ELEMENT: area.stack(position(summary.percent(age*COUNT, base.coordinate(dim(1)))),
    color.interior(sexd_p), missing.wings())
END GPL.

with graph

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=age MEAN(height)[name="MEAN_height"]
    MEAN(weight)[name="MEAN_weight"] MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: age=col(source(s), name("age"), unit.category())
  DATA: MEAN_height=col(source(s), name("MEAN_height"))
  DATA: MEAN_weight=col(source(s), name("MEAN_weight"))
  GUIDE: axis(dim(1), label("Возраст"))
  GUIDE: axis(scale(y1), label("Mean Длина тела, см"), color(color."3E58AC"))
  GUIDE: axis(scale(y2), label("Mean Масса тела, кг"), color(color."2EB848"), opposite())
  SCALE: cat(dim(1), include("10", "11", "12", "13", "14", "15", "16"))
  SCALE: y1 = linear(dim(2), include(0))
  SCALE: y2 = linear(dim(2), include(0))
  ELEMENT: line(position(age*MEAN_height), missing.wings(), color.interior(color."3E58AC"),
    scale(y1))
  ELEMENT: line(position(age*MEAN_weight), missing.wings(), color.interior(color."2EB848"),
    scale(y2))
END GPL.

I failed to combine. The problem is that both dual Y axes are occupied by HEIGHT and WEIGHT and so there no room
for Y axis serving for the Area element. Even though I actually don't need that percent Y axis to be displayed, it seemingly is necessary
as a gauge, and so there 3 Y axes in all - too many for GPL.

Can you help to solve it?

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

[Andreas Sprenger]

I don't know whether it makes sense to show plots with three y axes. I was interested in a possible graph but could not find a solution using ggraph. Therefore I used a vector graphics program and copied the graphs (right mouse click - copy special - metafile) into the graphics program. I matched the size of the graphs, removed the percent y-axis of graph 1. Subsequently I superimposed the two graphs. This is the result: GGraph_Example.pdf It makes sense to produce a single graph but if you like to produce graphs like this regularly it is too time consuming ;-) Cheers Andreas

---

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

[Andy W]

Here is your monstrosity Kirill,

<http://spssx-discussion.1045642.n5.nabble.com/file/t329824/Superimposed.png>

************************************************************************************************************************************.
DATA LIST FREE / age (F2.0) height (F3.0) weight (F3.2) sexd_p (F1.0).
BEGIN DATA
10      134     38.0        0
10      132     32.0        0
10      146     84.0        0
10      154     39.0        0
10      132     29.0        0
10      140     30.0        0
11      142     43.0        0
12      168     58.0        0
12      143     52.0        0
13      170     61.0        0
10      141     31.5        1
11      151     48.0        1
12      150     41.0        1
12      144     51.0        1
13      160     55.0        1
13      166     68.0        1
14      170     55.0        1
14      175     70.0        1
14      173     76.7        1
14      169     89.8        1
14      168     58.0        1
12      155     52.5        2
13      160     65.0        2
14      183     59.0        2
14      173     56.0        2
14      159     47.0        2
14      168     53.0        2
15      179     65.0        2
15      177     88.0        2
16      165     54.0        2
14      161     58.0        3
14      168     61.0        3
14      159     44.0        3
14      168     57.0        3
14      163     62.0        3
14      164     50.0        3
14      173     66.0        3
15      170     66.4        3
15      171     50.0        3
15      177     57.0        3
15      166     70.0        3
15      177     54.0        3
15      170     55.0        3
15      152     49.0        3
15      156     50.0        3
16      171     65.0        3
16      175     62.0        3
16      188     72.0        3
16      159     66.0        3
16      169     60.0        3
16      191     80.0        3
16      161     58.0        4
16      185     67.0        4
END DATA.
DATASET NAME Phys.
EXECUTE.

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=age COUNT()[name="COUNT"]
sexd_p MISSING=LISTWISE
    REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: age=col(source(s), name("age"), unit.category())
  DATA: COUNT=col(source(s), name("COUNT"))
  DATA: sexd_p=col(source(s), name("sexd_p"), unit.category())
  GUIDE: axis(dim(1), label("Возраст"))
  GUIDE: axis(dim(2), label("Percent"))
  GUIDE: legend(aesthetic(aesthetic.color.interior), label("P - лобковое
оволосение"))
  SCALE: cat(dim(1), include("10", "11", "12", "13", "14", "15", "16"))
  SCALE: linear(dim(2), include(0))
  SCALE: cat(aesthetic(aesthetic.color.interior), include("0", "1", "2",
"3", "4"))
  ELEMENT: area.stack(position(summary.percent(age*COUNT,
base.coordinate(dim(1)))),
    color.interior(sexd_p), missing.wings())
END GPL.

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=age
MEAN(height)[name="MEAN_height"]
    MEAN(weight)[name="MEAN_weight"] MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: age=col(source(s), name("age"), unit.category())
  DATA: MEAN_height=col(source(s), name("MEAN_height"))
  DATA: MEAN_weight=col(source(s), name("MEAN_weight"))
  GUIDE: axis(dim(1), label("Возраст"))
  GUIDE: axis(scale(y1), label("Mean Длина тела, см"),
color(color."3E58AC"))
  GUIDE: axis(scale(y2), label("Mean Масса тела, кг"),
color(color."2EB848"), opposite())
  SCALE: cat(dim(1), include("10", "11", "12", "13", "14", "15", "16"))
  SCALE: y1 = linear(dim(2), include(0))
  SCALE: y2 = linear(dim(2), include(0))
  ELEMENT: line(position(age*MEAN_height), missing.wings(),
color.interior(color."3E58AC"),
    scale(y1))
  ELEMENT: line(position(age*MEAN_weight), missing.wings(),
color.interior(color."2EB848"),
    scale(y2))
END GPL.

*Putting the two together.
COMPUTE sexd_p = sexd_p + 1.
VECTOR SDum(5,F1.0).
COMPUTE SDum(sexd_p) = 1.
EXECUTE.

DATASET DECLARE Agg.
AGGREGATE OUTFILE='Agg'
  /BREAK age
  /SDum1 TO SDum5 = SUM(SDum1 TO SDum5)
  /height = MEAN(height)
  /weight = MEAN(weight).

DATASET ACTIVATE Agg.
RECODE SDum1 TO SDum5 (SYSMIS = 0).
VARSTOCASES /MAKE CountSex FROM SDum1 TO SDum5 /INDEX sexd_p.
COMPUTE sexd_p = sexd_p - 1.

AGGREGATE OUTFILE=* MODE=ADDVARIABLES
  /BREAK Age
  /TotalSex = SUM(CountSex).


COMPUTE PropSex = CountSex/TotalSex.
*Need to put proportion of total within each category relative to some
external value.
*Here just choose 200.
COMPUTE RelSex = PropSex*200.
FORMATS Height Weight RelSex (F3.0).
DO IF age = LAG(age).
  COMPUTE height = $SYSMIS.
 COMPUTE weight = $SYSMIS.
END IF.
EXECUTE.

*Now putting all of these onto one chart.

GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=age height weight PropSex
sexd_p MISSING=VARIABLEWISE
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: age=col(source(s), name("age"), unit.category())
  DATA: height=col(source(s), name("height"))
  DATA: weight=col(source(s), name("weight"))
  DATA: sexd_p=col(source(s), name("sexd_p"), unit.category())
  DATA: PropSex=col(source(s), name("PropSex"))
  TRANS: RelSex = eval(PropSex*200)
  GUIDE: axis(dim(1), label("Возраст"))
  GUIDE: axis(scale(y1), label("Mean Длина тела, см"),
color(color."3E58AC"))
  GUIDE: axis(scale(y2), label("Mean Масса тела, кг"),
color(color."2EB848"), opposite())
  SCALE: cat(dim(1), include("10", "11", "12", "13", "14", "15", "16"))
  SCALE: y1 = linear(dim(2), include(0), max(200))
  SCALE: y2 = linear(dim(2), include(0))
  ELEMENT: area.stack(position(age*RelSex), color.interior(sexd_p),
missing.gap(), scale(y1))
  ELEMENT: line(position(age*height), missing.wings(),
color.interior(color."3E58AC"),
    scale(y1))
  ELEMENT: line(position(age*weight), missing.wings(),
color.interior(color."2EB848"),
    scale(y2))
END GPL.
************************************************************************************************************************************.



-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
Sent from: http://spssx-discussion.1045642.n5.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

[Kirill Orlov]

Andy, thank you very much for the help and forgive please for stealing your time. My request was indeed monstrosity, I admit it!

15.05.2018 18:23, Andy W пишет:

Here is your monstrosity Kirill, 

<http://spssx-discussion.1045642.n5.nabble.com/file/t329824/Superimposed.png> 

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

[PRogman]

Another possibility is to use multiple graphs, to get a second axis to the
right of the graph, explaining the proportions (I, too, agree on the
difficulty on interpreting the graph). With this solution there is no need
to rescale the proportions. In this example the secondary graph is empty,
just forcing the display of the axis. It is the width of the graph area that
moves the axis. The only post-production touch-up is to move the Color
Legend Box more to the right. The placement fits, IMHO, as the color scale
is related to the stacked area graph.
/PR

<http://spssx-discussion.1045642.n5.nabble.com/file/t339873/CombinedGraphs.png>

GGRAPH
  /GRAPHDATASET
   NAME      = "graphdataset"
   VARIABLES = age height weight PropSex sexd_p
   MISSING   = VARIABLEWISE
  /GRAPHSPEC
   SOURCE    = INLINE.
BEGIN GPL
  SOURCE:  s=userSource(id("graphdataset"))
  DATA:    age    =col(source(s), name("age"))
  DATA:    height =col(source(s), name("height"))
  DATA:    weight =col(source(s), name("weight"))
  DATA:    sexd_p =col(source(s), name("sexd_p"), unit.category())
  DATA:    PropSex=col(source(s), name("PropSex"))
  TRANS:   RelSex =eval(PropSex*100)
  TRANS:   Zero   =eval(0)

 GRAPH:   begin(origin(8%, 10%), scale(58%, 80%))
  GUIDE:   axis(dim(1)
               ,label("Age (yr)")
               )
  GUIDE:   axis(scale(y1)
               ,label("Height (cm)")
               ,delta(10)
               ,color(color."8c2d04")
               )
  GUIDE:   axis(scale(y2)
               ,label("Weight (kg)")
               ,delta(5)
               ,color(color."d94801")
               ,opposite()
               )
  GUIDE:   axis(scale(y3)
               ,label("%")
               ,delta(10)
               ,color(color.red)
               ,opposite()
               ,null()
               )
  GUIDE:   legend(aesthetic(aesthetic.color.interior)
                 ,label("Sexual development")
                 )
  SCALE:   cat(aesthetic(aesthetic.color.interior)
              ,include("0", "1", "2", "3", "4")
              ,map(("0",color."feedde")
                  ,("1",color."fdd0a2")
                  ,("2",color."fdae6b")
                  ,("3",color."fd8d3c")
                  ,("4",color."f16913")
                  ,("5",color."d94801")
                  ,("6",color."8c2d04")
                  )
              )
  SCALE:   linear(dim(1)
                 )
  SCALE:   y1 = linear(dim(2)
                      ,include(0)
                      ,max(200)
                      )
  SCALE:   y2 = linear(dim(2)
                      ,include(0)
                      )
  SCALE:   y3 = linear(dim(2)
                      ,include(0)
                      ,max(100)
                      )
  ELEMENT: area.stack(position(age*RelSex)
                     ,color.interior(sexd_p)
                     ,missing.gap()
                     ,scale(y3)
                     )
  ELEMENT: line(position(age*height)
               ,missing.wings()
               ,color.interior(color."8c2d04")
               ,scale(y1)
               )
  ELEMENT: line(position(age*weight)
               ,missing.wings()
               ,color.interior(color."d94801")
               ,scale(y2)
               )
 GRAPH:   end()

 GRAPH:   begin(origin(8%, 10%), scale(67%, 80%))
  GUIDE:   axis(dim(1)
               ,label("Age (yr)")
               ,null()
               )
  GUIDE:   axis(scale(y3)
               ,label("Percent")
               ,delta(10)
               ,color(color.red)
               ,opposite()
               )
  GUIDE:   legend(aesthetic(aesthetic.color.interior)
                 ,null()
                 )
  SCALE:   linear(dim(1)
                 )
  SCALE:   y3 = linear(dim(2)
                      ,include(0)
                      ,max(100)
                      )
  ELEMENT: area.stack(position(Zero*Zero)
                     ,color.interior(sexd_p)
                     ,missing.gap()
                     ,scale(y3)
                     )
 GRAPH:   end()
END GPL.




--
Sent from: http://spssx-discussion.1045642.n5.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

[Kirill Orlov]

PRogman, Andreas,
Thank you much for your answers. Both your answers were helpful, as Andrew's answer was.

16.05.2018 15:05, PRogman пишет:

Another possibility is to use multiple graphs, to get a second axis to the
right of the graph, explaining the proportions (I, too, agree on the
difficulty on interpreting the graph). With this solution there is no need
to rescale the proportions. In this example the secondary graph is empty,
just forcing the display of the axis. It is the width of the graph area that
moves the axis. The only post-production touch-up is to move the Color
Legend Box more to the right. The placement fits, IMHO, as the color scale
is related to the stacked area graph.
/PR

<http://spssx-discussion.1045642.n5.nabble.com/file/t339873/CombinedGraphs.png> 

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