How can I specify which columns to select using read.table in R

0
votes

I have a dataset with 100 columns and it doesn't have a header.

I have an int vector that consists of some numbers ranges between 1 to 100. For example, a vector with "2 5 62 78".

Now when I read the dataset using read.table, all I want is to select column 2, 5, 62 and 78 from the dataset. How can I do that? Many thanks.

1
rread.table
Maybe this will help you? stackoverflow.com/questions/16320667/… – Christian Borck
Take a look at ?read.table option colClasses. If you set it to "NULL" (that's a character), it will skip the column. – ilir
@ilir you should post that as the answer. Tho' I suspect for a 100-column file, it'll be much easier to do foo<-read.table(datafile); foo<-foo[,c(2,5,62,78)] than foo<-read.table(datafile,colClasses=c(rep('numeric',1),NULL,rep('numeric',2),...) – Carl Witthoft
@CarlWitthoft you can populate the vector of classes beforehand, and use some tricks. See my answer for something I've had to use a lot in the past. – ilir
@ilir Yep, I upvoted on account of your detailed approach. – Carl Witthoft

1 Answers

3
votes

What you want is the option colClasses of read.table() (and the derivative functions). It allows you to pass a character vector with the classes of each column in the data. If you set that to "NULL" the column will be skipped. You can set the whole thing to "NULL" and then only change the ones you want to import (based on their class).

Proof of concept below.

cc <- rep('NULL', 100)       ## skip all 100 columns
cc[c(2, 5)] <- 'integer'     ## 2 and 5 are integer
cc[c(62, 58)] <- 'character' ## 62 and 58 will be imported as character
df <- read.csv('really-wide-data.csv', colClasses=cc)