help advanced_options
-------------------------------------------------------------------------------
Title
[G] advanced_options -- Rarely specified options for use with graph
twoway
Syntax
title_options description
---------------------------------------------------------------------
pcycle(#) plots before pstyles recycle
yvarlabel(quoted_strings) respecify y-variable labels
xvarlabel(quoted_string) respecify x-variable label
yvarformat(%fmt [...]) respecify y-variable formats
xvarformat(%fmt) respecify x-variable format
yoverhangs adjust margins for y-axis labels
xoverhangs adjust margins for x-axis labels
recast(newplottype) treat plot as newplottype
---------------------------------------------------------------------
The above options are rightmost; see repeated options.
where quoted_string is one quoted string and quoted_strings are one or
more quoted strings, such as
"plot 1 label"
"plot 1 label" "plot 2 label"
newplottype description
---------------------------------------------------------------------
scatter treat as graph twoway scatter
line treat as graph twoway line
connected treat as graph twoway connected
bar treat as graph twoway bar
area treat as graph twoway area
spike treat as graph twoway spike
dropline treat as graph twoway dropline
dot treat as graph twoway dot
rarea treat as graph twoway rarea
rbar treat as graph twoway rbar
rspike treat as graph twoway rspike
rcap treat as graph twoway rcap
rcapsym treat as graph twoway rcapsym
rline treat as graph twoway rline
rconnected treat as graph twoway rconnected
rscatter treat as graph twoway rscatter
pcspike treat as graph twoway pcspike
pccapsym treat as graph twoway pccapsym
pcarrow treat as graph twoway pcarrow
pcbarrow treat as graph twoway pcbarrow
pcscatter treat as graph twoway pcscatter
---------------------------------------------------------------------
newplottypes in each grouping (scatter through dot, rarea though
rscatter, and pcspike through pcscatter) should be recast only among
themselves.
Description
The advanced_options are not so much advanced as they are difficult to
explain and are rarely used. They are also invaluable when you need
them.
Options
pcycle(#) specifies how many plots are drawn before the pstyle of the
next plot begins again at p1, with the plot after the next plot using
p2, and so on. The default # for most schemes is 15.
yvarlabel(quoted_strings) and xvarlabel(quoted_string) specify strings
that are to be treated as if they were the variable labels of the
first, second, ..., y variables and of the x variable.
yvarformat(%fmt) and xvarformat(%fmt) specify display formats that are to
be treated as if they were the display formats of the first, second,
..., y variables and of the x variable.
yoverhangs and xoverhangs attempt to adjust the graph region margins to
prevent long labels on the y or x axis from extending off the edges
of the graph. Only the labels for the smallest and largest tick
values on the axes are considered when making the adjustment.
yoverhangs and xoverhangs are ignored if by() is specified.
recast(newplottype) specifies the new plottype to which the original
graph twoway plottype command is to be recast.
Remarks
Remarks are presented under the following headings:
Use of yvarlabel() and xvarlabel()
Use of yvarformat() and xvarformat()
Use of recast()
Use of yvarlabel() and xvarlabel()
When you type, for instance,
. scatter mpg weight
the axes are titled using the variable labels of mpg and weight or, if
the variables have no variable labels, using the names of the variables
themselves. Options yvarlabel() and xvarlabel() allow you to specify
strings that will be used in preference to both the variable label and
the name.
. scatter mpg weight, yvarl("Miles per gallon")
would label the y axis "Miles per gallon" (omitting the quotes),
regardless of how variable mpg was labeled. Similarly,
. scatter mpg weight, xvarl("Weight in pounds")
would label the x axis "Weight in pounds", regardless of how variable
weight was labeled.
Obviously, you could specify both options.
In neither case will the actual variable label be changed. Options
yvarlabel() and xvarlabel() treat the specified strings as if they were
the variable labels. yvarlabel() and xvarlabel() are literal in this
treatment. If you specified xvarlabel(""), for instance, the variable
label would be treated as if it were nonexistent, and thus the variable
name would be used to title the x axis.
What makes these two options "advanced" is not only that they affect the
way axes are titled but also that they substitute the specified strings
for the variable labels wherever the variable label might be used.
Variable labels are also used, for instance, in the construction of
legends (see [G] legend_option).
Use of yvarformat() and xvarformat()
Options yvarformat() and xvarformat() work much like yvarlabel() and
xvarlabel(), except that, rather than overriding the variable labels,
they override the variable formats. If you type
. scatter mpg weight, yvarformat(%9.2f)
the values on the y axis will be labeled 10.00, 20.00, 30.00, and 40.00
rather than 10, 20, 30, and 40.
Use of recast()
scatter, line, histogram, ... -- the word that appears directly after
graph twoway -- is called a plottype. Plottypes come in two forms: base
plottypes and derived plottypes.
Base plottypes plot the data as given according to some style. scatter
and line are examples of base plottypes.
Derived plottypes do not plot the data as given but instead derive
something from the data and then plot that according to one of the base
plottypes. histogram is an example of a derived plottype. It derives
from the data the values for the frequencies at certain x ranges, and
then it plots that derived data using the base plottype graph twoway bar.
lfit is another example of a derived plottype. It takes the data, fits a
linear regression, and then passes that result along to graph twoway
line.
recast() is useful when using derived plottypes. It specifies that the
data are to be derived just as they would be ordinarily, but rather than
passing the derived data to the default base plottype for plotting, they
are passed to the specified base plottype.
For instance, if we typed
. twoway lfit mpg weight, pred(resid)
we would obtain a graph of the residuals as a line plot because the lfit
plottype produces line plots. If we typed
. twoway lfit mpg weight, pred(resid) recast(scatter)
we would obtain a scatterplot of the residuals. graph twoway lfit would
use graph twoway scatter rather than graph twoway line to plot the data
it derives.
recast(newplottype) may be used with both derived and base plottypes,
although it is most useful when combined with derived plots.
Technical note:
The syntax diagram shown for scatter in [G] graph twoway scatter,
although extensive, is incomplete, and so are all the other plottype
syntax diagrams shown in this manual.
Consider what would happen if you specified
. scatter ..., ... recast(bar)
You would be specifying that scatter be treated as a bar. Results
would be the same as if you typed
. twoway bar ..., ...
but let's ignore that and pretend that you typed the recast()
version. What if you wanted to specify the look of the bars? You
could type
. scatter ..., ... bar_options recast(bar)
That is, scatter allows graph twoway bar's options, even though they
do not appear in scatter's syntax diagram. Similarly, graph twoway
bar allows all of scatter's options; you might type
. twoway bar ..., ... scatter_options recast(scatter)
The same is true for all other pairs of base plottypes, with the
result that all base plottypes allow all base plottype options. The
emphasis here is on base: the derived plottypes do not allow this
sharing.
If you use a base plottype without recast() and if you specify
irrelevant options from other base types, that is not an error, but
the irrelevant options are ignored. In the syntax diagrams for the
base plottypes, we have listed only the options that matter under the
assumption that you do not specify recast.
++
++
Also see
Manual: [G] advanced_options
Help: [G] graph twoway
