Hi,
I'm ccing your message to the MatchIt mailserve to which any questions
should be directed.
The memory problem is not specific to MatchIt, but is a general issue
using R. If you are using a Windows machine, I believe that you can
follow the instruction given here;
http://gking.harvard.edu/zelig/docs/static/faq.pdf
If you are using a 64 bit Mac/Unix machine, then you might consider
optimizing R. I did it following http://r.research.att.com/ and it worked
well for me.
The last resort, which might be the easiest, would be to subset the data
by matching exactly on the discrete covariate and then do further matching
within that subset of the data.
Best,
Kosuke
--
Department of Politics
Princeton University
http://imai.princeton.edu
On Mon, 16 Mar 2009, Farshad Farzadfar wrote:
> Dear Dr. Imai,
>
> I'm a doctoral student in Harvard School of Public Health and using MatchIt package (full match) to estimate the effect of treatment. I have a dataset with 15000 observations and 10 variables. When I'm running matchIt commands it gives me this message: "cannot allocate vector of size 455.5 Mb". I used memory.size (4000) to optimize the memory but still have problem. I changed my computer to another one with 8GB RAM and 1.8 MHtz processor, but the same result. I would be pleased if you let me know how I can deal with this problem.
>
> Best Regards
>
> Farshad Farzadfar
>
>
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
Howdy,
I have a question regarding NN matching. My call looks as follows:
matchit(match.f.2,p.2,method="nearest",distance="logit",ratio=1,caliper=.1)->m.nc.2
In the documentation, the m.order defaults to "largest" so the treated
units are paired up from the highest value of the distance to the
lowest. I've run this a number of times and gotten different results
every time; my first thought was that the controls were randomly
arranged each time such that a different match is being made. Looking
at the matchit2nearest source (which I haven't completely deciphered)
makes me not so sure about that idea. What is going on here? And is
there a way to "fix" this option such that the same match is made
every time (specifically, it's making a certain sweave document
difficult to write when things keep changing).
Thanks for any advice,
Ben
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/