Question

在阅读下面的文本时，fread()无法检测第8列和第9列中的缺失值。这仅使用默认选项integer64="integer64"。设置integer64="double"或"character"会正确检测NA。请注意，该文件在V8和V9中有三种可能的NA-- ,,; , ,;和NA。附加na.strings=c("NA","N/A",""," "), sep=","作为选项无效。

使用read.csv()的方式与fread(integer="double")的工作方式相同。

要阅读的文字（也是available as a file integer64_and_NA.csv）：

2012,276,,0,"S1","001",1,,724135215,1590915056,
2012,276,2,8,"S1","001",1, ,,154598,0
2012,276,2,12,"S1","001",1,NA,5118863,21819477,
2012,276,2,0,"S1","011",8,3127133583,3127133583,9003982501,0

以下是fread()的输出：

DT <- fread(input="integer64_and_NA.csv", verbose=TRUE, integer64="integer64", na.strings=c("NA","N/A",""," "), sep=",")

Input contains no \n. Taking this to be a filename to open
Detected eol as \r\n (CRLF) in that order, the Windows standard.
Looking for supplied sep ',' on line 4 (the last non blank line in the first 'autostart') ... found ok
Found 11 columns
First row with 11 fields occurs on line 1 (either column names or first row of data)
Some fields on line 1 are not type character (or are empty). Treating as a data row and using default column names.
Count of eol after first data row: 5
Subtracted 1 for last eol and any trailing empty lines, leaving 4 data rows
Type codes: 11114412221 (first 5 rows)
Type codes: 11114412221 (after applying colClasses and integer64)
Type codes: 11114412221 (after applying drop or select (if supplied)
Allocating 11 column slots (11 - 0 NULL)
   0.000s (  0%) Memory map (rerun may be quicker)
   0.000s (  0%) sep and header detection
   0.000s (  0%) Count rows (wc -l)
   0.000s (  0%) Column type detection (first, middle and last 5 rows)
   0.000s (  0%) Allocation of 4x11 result (xMB) in RAM
   0.000s (  0%) Reading data
   0.000s (  0%) Allocation for type bumps (if any), including gc time if triggered
   0.000s (  0%) Coercing data already read in type bumps (if any)
   0.000s (  0%) Changing na.strings to NA
   0.001s        Total

结果data.table是：

DT
     V1  V2 V3 V4 V5  V6 V7                  V8                  V9        V10 V11
1: 2012 276 NA  0 S1 001  1 9218868437227407266           724135215 1590915056  NA
2: 2012 276  2  8 S1 001  1 9218868437227407266 9218868437227407266     154598   0
3: 2012 276  2 12 S1 001  1 9218868437227407266             5118863   21819477  NA
4: 2012 276  2  0 S1 011  8          3127133583          3127133583 9003982501   0

在非NA的列中正确检测到

integer64个值。对于V8和V9，fread()标记为整数64，而不是NA，我们有“9218868437227407266”。有趣的是，str()将V8和V9的相应值作为NA返回：

str(DT)

Classes ‘data.table’ and 'data.frame':  4 obs. of  11 variables:
 $ V1 : int  2012 2012 2012 2012
 $ V2 : int  276 276 276 276
 $ V3 : int  NA 2 2 2
 $ V4 : int  0 8 12 0
 $ V5 : chr  "S1" "S1" "S1" "S1"
 $ V6 : chr  "001" "001" "001" "011"
 $ V7 : int  1 1 1 8
 $ V8 :Class 'integer64'  num [1:4] NA NA NA 1.55e-314
 $ V9 :Class 'integer64'  num [1:4] 3.58e-315 NA 2.53e-317 1.55e-314
 $ V10:Class 'integer64'  num [1:4] 7.86e-315 7.64e-319 1.08e-316 4.45e-314
 $ V11: int  NA 0 NA 0
 - attr(*, ".internal.selfref")=<externalptr>

...但没有其他任何内容将其视为NA：

is.na(DT$V8)
[1] FALSE FALSE FALSE FALSE
max(DT$V8)
integer64
[1] 9218868437227407266
> max(DT$V8, na.rm=TRUE)
integer64
[1] 9218868437227407266
> class(DT$V8)
[1] "integer64"
> typeof(DT$V8)
[1] "double"

它似乎不仅仅是打印/屏幕问题，data.table将它们视为巨大的整数：

DT[, V12:=as.numeric(V8)]
Warning message:
In as.double.integer64(V8) :
  integer precision lost while converting to double
> DT
     V1  V2 V3 V4 V5  V6 V7                  V8                  V9        V10 V11          V12
1: 2012 276 NA  0 S1 001  1 9218868437227407266           724135215 1590915056  NA 9.218868e+18
2: 2012 276  2  8 S1 001  1 9218868437227407266 9218868437227407266     154598   0 9.218868e+18
3: 2012 276  2 12 S1 001  1 9218868437227407266             5118863   21819477  NA 9.218868e+18
4: 2012 276  2  0 S1 011  8          3127133583          3127133583 9003982501   0 3.127134e+09

我是否遗漏了integer64的内容，或者这是一个错误？如上所述，我可以使用integer64="double"，可能会失去一些精度，如帮助文件中所述。但意外的行为是使用默认的integer64 ...

这是在运行Revolution R 3.0.2的Windows 8.1 64位计算机上运行的，也是在运行kubuntu 13.10，CRAN-R 3.0.2的虚拟机上完成的。使用CRAN（2014年2月7日1.8.10）和1.8.11（rev.1110,2014-02-04 02:43:19）的最新稳定数据表进行测试，从拉链手动安装为r-forge在Windows上只有稳定的1.8.10，而且只有稳定的1.8.10。在两台机器上安装并加载了bit64。

> sessionInfo()
R version 3.0.2 (2013-09-25)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252    LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] grid      stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] bit64_0.9-3       bit_1.1-11        gdata_2.13.2      xts_0.9-7         zoo_1.7-10        nlme_3.1-113      hexbin_1.26.3     lattice_0.20-24   ggplot2_0.9.3.1  
[10] plyr_1.8          reshape2_1.2.2    data.table_1.8.11 Revobase_7.0.0    RevoMods_7.0.0    RevoScaleR_7.0.0 

loaded via a namespace (and not attached):
 [1] codetools_0.2-8    colorspace_1.2-4   dichromat_2.0-0    digest_0.6.4       foreach_1.4.1      gtable_0.1.2       gtools_3.2.1       iterators_1.0.6   
 [9] labeling_0.2       MASS_7.3-29        munsell_0.4.2      proto_0.3-10       RColorBrewer_1.0-5 reshape_0.8.4      scales_0.2.3       stringr_0.6.2     
[17] tools_3.0.2

Answer 1

此错误#488现在已在data.table v1.9.5的开发版本中使用this commit修复，并且值已正确分配（并显示）为NA bit64已加载。

require(data.table) # v1.9.5
require(bit64)
ans = fread("test.csv")
#      V1  V2 V3 V4 V5  V6 V7         V8         V9        V10 V11
# 1: 2012 276 NA  0 S1 001  1         NA  724135215 1590915056  NA
# 2: 2012 276  2  8 S1 001  1         NA         NA     154598   0
# 3: 2012 276  2 12 S1 001  1         NA    5118863   21819477  NA
# 4: 2012 276  2  0 S1 011  8 3127133583 3127133583 9003982501   0

Answer 2

这显然是bit64包的问题，而不是fread()或data.table。来自bit64文档http://cran.r-project.org/web/packages/bit64/bit64.pdf

“目前不支持订阅不存在的元素和订阅NAs。这样的订阅当前返回9218868437227407266而不是NA（非代数双重代码的NA值）。遵循完整的R行为会破坏性能或需要大量的C编码。“

我尝试将9218868437227407266的值重新分配给NA认为它会起作用

实施例。

DT[V8==9218868437227407266, ]
#actually returns nothing, but
DT[V8==max(V8), ]
#returns the rows with 9218868437227407266 in V8
#but this does not reassign the value 
DT[V8==max(V8), V8:=NA]
#not that this makes sense, but I tried just in case...
DT[V8==max(V8), V8:=NA_character_]

因此，文档非常清楚地说明，如果向量是类integer64，它将无法识别NA或缺少值。我要避免bit64只是为了不必处理这个......

fread（）失败，在integer64列中缺少值

2 个答案: