fread in R imports a large .csv file as a data frame with one row

问题

I'm importing a large .csv file into R (about 0.5 million rows), so I've been trying to use fread() from the data.table package as a faster alternative to read.table() and read.csv(). However, fread() returns a data frame with all of the data from the rows inside one row, even though it has the correct number of columns. I found a bug report from 2013 showing this is related to the integer64 data class:

http://r-forge.r-project.org/tracker/index.php?func=detail&aid=2786&group_id=240&atid=975

Are there any fixes or ways to get around this?

The .csv file I'm trying to read is entirely integers ranging from 0 - 10000, with no missing data. I'm using R version 2.15.2 on a Windows 7 computer, with version 1.8.8 of the data.table package.

The code I'm running is:

require(data.table)
fread("pre2012_alldatapoints.csv", sep = ",", header= TRUE)-> pre
head(pre)

1: 1 22 -105 22 -105
2: 2 22 -105 22 -105
3: 3 20 -105 20 -105
4: 4 21 -105 21 -105
5: 5 21 -105 21 -105
6: 6 21 -105 21 -105

dim(pre)
[1] 12299  5 #dim returns the correct number of dimensions
#this is a subset of the file I want to import that I've confirmed imports correctly with read.csv

pre[,1]
[1] 1 #but trying to print a column returns this

length(pre[,1])
[1] 1 #and length for any column returns a row length of 1

Thanks a lot for your help!

回答1:

fread creates a data.table. The data.table package comes with a number of vignettes.

Your precise issues is addressed in FAQ 1.1 from the data.table FAQ - the very first FAQ!

By default the second argurment to [.data.table is an expression evaluated within the scope of the data.table

therefore pre[,1] evaluates 1 within the scope of pre. 1 is still 1. If you want' to reference by column number, use with=FALSE pre[,1,with=FALSE]

来源：https://stackoverflow.com/questions/22316251/fread-in-r-imports-a-large-csv-file-as-a-data-frame-with-one-row