<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 5.5.2653.12">
<TITLE>RE: [R] large dataframes to ascii</TITLE>
</HEAD>
<BODY>
<P><FONT FACE="Verdana">Hello,</FONT>
</P>
<P><FONT FACE="Verdana">One alternative way to do that is by processing portions of your big table.</FONT>
</P>
<P><FONT FACE="Verdana">You can select, consecutive numbers of colums and rows and save them in separate files that after finishing you can recompose on Linux with cat and paste commands.</FONT></P>
<P><FONT FACE="Verdana">It is clearly more tedious but effective and manageable.</FONT>
</P>
<P><FONT FACE="Verdana">Hope it helps.</FONT>
<BR><FONT FACE="Verdana">Carlos Ortega.</FONT>
</P>
<P><FONT FACE="Verdana">--------------------------------------------------------------------</FONT>
<BR><FONT FACE="Times New Roman">Hi R-users, </FONT>
<BR><FONT FACE="Times New Roman">I want to convert a large dataset (from stata format) to an ascii table.<BR>
The resulting table should be a human-readable table (like CSV, or<BR>
tab-separated file or something similar). R reads the stata-file quite<BR>
easily (with some problems which are discussed here earlier), but so long I<BR>
have not found a suitable way to write it in ascii format. </FONT>
<BR><FONT FACE="Times New Roman">Sure, there exists write.table, which is intended for that purpose. But the<BR>
problem is that the table is quite a large (7500 obs x 1200 vars) and after<BR>
half-an-hout processing on my PII wiht 128 MB memory, write.table stopped<BR>
with a message that it could not allocate 35 MB more. </FONT>
<BR><FONT FACE="Times New Roman">dput() and save( ..., ascii=TRUE) work well but the result is not such a<BR>
table I am looking for. I tried also print() with option( width=very wide)<BR>
but the largest allowed with is 10000, the table should be around 30000<BR>
characters wide. </FONT>
<BR><FONT FACE="Times New Roman">When I tried to write.table() line-by line, I got quite soon a message that<BR>
R was killed. </FONT>
<BR><FONT FACE="Times New Roman">I know the write.table() pastes together all the variables and writes then<BR>
the resulting string to the file. Aren't there any more efficient way of<BR>
doing it, I think saving the dataframe variable-by-variable using a C-function<BR>
should be quite a fast and not memory hungry at all. </FONT>
<BR><FONT FACE="Times New Roman">I have R 1.3.1 on RH7.1 linux </FONT>
<BR><FONT FACE="Times New Roman">Thanks in advance, </FONT>
<BR><FONT FACE="Times New Roman">Ott Toomet </FONT>
</P>
</BODY>
</HTML>