Here is some minor trickery which may be useful
to those who draw graphs of time series.
In spirit it is close to
http://www.stata.com/support/faqs/graphics/nicedate.html
but the problem is different and the solution is for
Stata 8.
I was playing with about 10 years' worth of
environmental data: the variables are for various
daily dates and all show strong seasonality. I wanted
therefore to show both the years "1987 ... 1997" and
the year ends. That is more labels than might otherwise
be justified aesthetically, but often with this kind of
data you want the detail for look-up, as you
can bring in other knowledge ("1993 was a very dry year,
so that peak makes sense", etc.). No doubt much the
same could be said of many analyses of economic or
other time series.
This loop puts all the 1 January dates and the 1 July
dates into local macros:
forval y = 1987/1997 {
local jan "`jan' `=mdy(1,1,`y')'"
local jul "`jul' `=mdy(7,1,`y')'"
}
Then the -graph- call is say
. twoway connected <whatever> <date>,
xla(`jul', format(%dCy) noticks) xtic(`jan', grid)
That is, text labels appear at 1 July (close enough to
the middle of the year for graphical purposes) formatted
so that only the years are shown, but with ticks
suppressed, while ticks and a grid appear at year ends.
Clearly this could be extended in various ways, e.g.
to non-calendar years, and it could be automated.
Just make sure that the graph code can see the local
macros. (For example, the graph code in a -doedit-
window cannot see the local macros in your main
interactive session.)
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
