Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: How to vertically align graphs with Stata 12 ?


From   Nick Cox <[email protected]>
To   [email protected]
Subject   Re: st: How to vertically align graphs with Stata 12 ?
Date   Mon, 13 Feb 2012 08:37:20 +0000

Thanks. That is now clear. The easiest way to solve this that I know
is to insist on -yla(, ang(v))- in both graphs. Then you can also add
-fysize()- for both graphs and -xsc(off)- for the first as below. Tune
-fysize()- to choice.

clear
set obs 200
egen week = seq(), to(100) block(2)
egen type = seq(), to(2)
gen DUM = runiform()
gen Y = exp(rnormal(1,1))
line DUM week, scheme(lean1) name(choice, replace) sort yla(, ang(v))
xsc(off) fysize(45)
twoway (line Y week if type==1,sort  lcolor(black)) ///
(line Y week if type==2,sort  lcolor(red)) ,  ///
scheme(lean1) name(graph, replace) legend(off) yla(, ang(v)) fysize(55)

graph combine choice graph, cols(1) xcommon

Nick

On Sun, Feb 12, 2012 at 4:49 PM, Francesco <[email protected]> wrote:

> I have a panel database that I aggregate using the -collapse function.
> First of all, I generate a dummy variable that I would like to study (DUM)
>
> - gen DUM=variable_of_interest==1
>
> then I collapse the database by week and type of individuals (2 types,
> identified by the dummy variable TYPE)
>
> -collapse (count) count_Y=Y  ///
> (mean) mean_Y=Y  ///
> (mean) mean_DUM=DUM  ///
> , by(week type) fast
>
> I create the first graph, which requires at leat 10 datapoints per
> week and plots the evolution of the DUM's proportion for type 1
> individuals
>
> - line mean_DUM week if count_Y>10&`TYPE'==1 , scheme(lean1)
> name(choice, replace) sort
>
> And I would like to plot it against the output Y for the two different types :
>
> -twoway (line mean_Y week if count_Y>10&`TYPE'==1,sort  lcolor(black))
> (line mean_Y week if count_Y>10&`TYPE'==0,sort  lcolor(red)) ,
> scheme(lean1) name(graph, replace) legend(off)
>
> Therefore I type :
> -graph combine choice graph, cols(1) xcommon
>
> The graph that I obtain is exactly what I want (one graph per row)...
> except for the (important) fact that there is a slight lag in the
> alignment of the x-axis due to (I guess) the difference in length of
> digits between DUM (0.2 for example) and Y (0.002 for example) ...


> On 12 February 2012 15:13, Nick Cox <[email protected]> wrote:

>> The thread you cite from 2003 concerns graphs from -graph hbar-. In
>> that case the graphs are rather different from yours given Stata's use
>> of categorical axes.
>>
>> As you have results from -line-, meaning -twoway line-, I would expect
>> -graph combine- to give good results but it may require use of the
>> -xcommon- option.
>>
>>
>> On Sun, Feb 12, 2012 at 12:15 PM, Francesco <[email protected]> wrote:
>>
>>> I created 3 (sub)graphs using the -line command, where the x-axis
>>> represents dates (months, years, etc), and I would like to align them
>>> vertically.
>>> That is obtaining one single graphical output that displays one
>>> (sub)graph per row and with the following important feature : if you
>>> draw an imaginary vertical line from a particular date of any graph
>>> you should cut the x-axis of the other graphs at the same date...
>>>
>>> I tried with the -graph combine option, but it does not work properly
>>> : there is a slight time lag probably due to y-axis labels... so the
>>> graphs are not perfectly aligned..
>>> Also I found a thread which basically concludes that it is not obvious
>>> to do this with Stata in a slightly different context :
>>> (http://www.stata.com/statalist/archive/2003-12/msg00329.html)
>>>
>>> However this post was made in ... 2003 ! I am sure that some
>>> improvements have been introduced since...

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index