[R] Saving tab/csv delimited data with NaN's

[Rolf Turner]

On 24/03/2010, at 12:34 PM, Sharpie wrote:

> 
> 
> shankar-17 wrote:
>> 
>> Hello,
>> I am working multiple simulated data sets with missing values, I would  
>> like to store these data sets in either tab delimited format for .csv  
>> format with missing values marked as NaN's instead of NA's.
>> 
>> I read the import/export document which mentions that write.table  
>> command converts NaN's to NA. Is there any other way I can store the  
>> NaN's. I tried the write syntax it gives me error codes.
>> Each data files are of dimensions 1000 x 21 .
>> 
>> I would appreciate any help in this regard.
>> 
>> Many thanks
>> 
> 
>> foo <- matrix(0,nrow=3,ncol=3)
>> foo
>     [,1] [,2] [,3]
> [1,]    0    0    0
> [2,]    0    0    0
> [3,]    0    0    0
> 
>> foo[3,3] <- NA
>> foo
>     [,1] [,2] [,3]
> [1,]    0    0    0
> [2,]    0    0    0
> [3,]    0    0   NA
> 
>> write.csv( foo, file='tst.csv', na = "NaN", row.names = F )
>> readLines( 'tst.csv' )
> [1] "\"V1\",\"V2\",\"V3\"" "0,0,0"                "0,0,0"
> [4] "0,0,NaN"
> 
> Seems to work fine for me.  If you post a reproducible example, we could
> probably figure out why it is not working for you.


This doesn't work if you have both NAs and NaNs in your data frame and you
want to distinguish between these.  I.e. when you read the data back in,
all NAs will have been converted to NaNs.

Admittedly the OP said he wanted to represent all NAs as NaNs, so your
solution would seem to work for him.

	cheers,

		Rolf Turner

######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}