[R] foreign library - negative integers??

[Thomas Lumley]

On Thu, 16 May 2002, Michaell Taylor wrote:

>
> I am having a problem with the foreign library correctly reading some integer
> data.  Specifically,
>
> d _ read.dta('aptaa.dta')
> > d[1:5,]
>   scenario metcode   yr  ginv cons   gocc   abs   dvac   gmre   gmer
> 1        1      AA 2002 0.007 1377 -0.071 51710  0.071 -0.011 -0.127
> 2        1      AA 2003 0.000    0 -0.016 62568  0.014 -0.043 -0.538
> 3        1      AA 2004 0.000    0 -0.002 65122  0.002 -0.090 -0.338
> 4        1      AA 2005 0.000    0  0.000 65528  0.000 -0.036 -0.272
> 5        1      AA 2006 0.000    0  0.002   309 -0.001 -0.050  0.468
> > dd _ read.csv('aptaa.csv',header=T)
> > dd[1:5,]
>   scenario metcode   yr  ginv cons   gocc    abs   dvac   gmre   gmer
> 1        1      AA 2002 0.007 1377 -0.071 -13826  0.071 -0.011 -0.127
> 2        1      AA 2003 0.000    0 -0.016  -2968  0.014 -0.043 -0.538
> 3        1      AA 2004 0.000    0 -0.002   -414  0.002 -0.090 -0.338
> 4        1      AA 2005 0.000    0  0.000     -8  0.000 -0.036 -0.272
> 5        1      AA 2006 0.000    0  0.002    309 -0.001 -0.050  0.468
>
> In theory, dd== d, but notice the differences in abs.
>
> The problem is that aptaa.csv is simply an "outsheet using" (in stata) version
> of aptaa.dta - and thus identical.  I have checked these two files (numerous
> times.).  (loading/reloading/rewriting in stata)
>
> It appears that the negative "abs" is a problem, but other variables are not
> suffering the same problem. (no other integer variables take negatives)
>

This looks like a signed vs unsigned integer problem. If so, a simple
work-around is to subtract 65536 from values greater than 32767.

abs <- ifelse(abs>2^15, abs-2^16, abs)

This also suggests that the Stata command
   recast long abs
might well be a more elegant work-around.


	-thomas


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._