堆栈内存溢出
  简体   繁体   English

data.table - 拆分多列

[英]data.table - split multiple columns

 原文  2014-12-31 11:43:04  6   2    r/ data.table

问题描述

I have a data.table G as follows.我有一个 data.table G如下。

d <- list( c("SD1:LUSH", "SD44:CANCEL", "SD384:FR563", "SD32:TRUMPET"), c("SD23:SWITCH", "SD1:LUSH", "SD567:TREK"), c("SD42:CRAYON", "SD345:FOX", "SD183:WIRE"), c("SD345:HOLE", "SD340:DUST", "SD387:ROLL"), c("SD455:TOMATO", "SD86:RAY", "SD39:MATURE", "SD86:COSMIC"), c("SD12:PAINTING", "SD315:MONEY31", "SD387:SPRING"),  c("SD32:TRUMPET", "SD1:FIELD", "SD40:GREEN", "SD40:PARK"))

d2 <-  lapply(d, function(x) sapply(strsplit(x, ":"), "[", 1))
d <- lapply(d, function(x) paste0(unique(x), collapse=", "))
d2 <- lapply(d2, function(x) paste0(unique(x), collapse=", "))
d <- as.data.frame(as.matrix(lapply(d, paste0, collapse=", ")))
d2 <- as.data.frame(as.matrix(lapply(d2, paste0, collapse=", ")))
d <- as.data.frame(cbind(d,d2))
colnames(d) <- c("sdw", "sd")
d$sd <- as.character(d$sd)
d$sdw <- as.character(d$sdw)

G <- data.table( d , key = "sd" )
G[, GNO:= seq(1,nrow(G))]
setcolorder(G, neworder=c("GNO", "sd", "sdw"))

G
    GNO                     sd                                              sdw
1:   1 SD1, SD44, SD384, SD32 SD1:LUSH, SD44:CANCEL, SD384:FR563, SD32:TRUMPET
2:   2     SD12, SD315, SD387       SD12:PAINTING, SD315:MONEY31, SD387:SPRING
3:   3       SD23, SD1, SD567                SD23:SWITCH, SD1:LUSH, SD567:TREK
4:   4        SD32, SD1, SD40   SD32:TRUMPET, SD1:FIELD, SD40:GREEN, SD40:PARK
5:   5    SD345, SD340, SD387               SD345:HOLE, SD340:DUST, SD387:ROLL
6:   6     SD42, SD345, SD183               SD42:CRAYON, SD345:FOX, SD183:WIRE
7:   7      SD455, SD86, SD39 SD455:TOMATO, SD86:RAY, SD39:MATURE, SD86:COSMIC

I am trying to split each set of elements in column sd and get a modified data.table according to column GNO as follows.我正在尝试拆分列sd每组元素，并根据列GNO获取修改后的GNO ，如下所示。

G2 <- G[, list(sd = unlist(strsplit( sd , "," ))) , by = list(GNO, sdw)]

G2
   GNO                                              sdw     sd
 1:   1 SD1:LUSH, SD44:CANCEL, SD384:FR563, SD32:TRUMPET    SD1
 2:   1 SD1:LUSH, SD44:CANCEL, SD384:FR563, SD32:TRUMPET   SD44
 3:   1 SD1:LUSH, SD44:CANCEL, SD384:FR563, SD32:TRUMPET  SD384
 4:   1 SD1:LUSH, SD44:CANCEL, SD384:FR563, SD32:TRUMPET   SD32
 5:   2       SD12:PAINTING, SD315:MONEY31, SD387:SPRING   SD12
 6:   2       SD12:PAINTING, SD315:MONEY31, SD387:SPRING  SD315
 7:   2       SD12:PAINTING, SD315:MONEY31, SD387:SPRING  SD387
 8:   3                SD23:SWITCH, SD1:LUSH, SD567:TREK   SD23
 9:   3                SD23:SWITCH, SD1:LUSH, SD567:TREK    SD1
10:   3                SD23:SWITCH, SD1:LUSH, SD567:TREK  SD567
11:   4   SD32:TRUMPET, SD1:FIELD, SD40:GREEN, SD40:PARK   SD32
12:   4   SD32:TRUMPET, SD1:FIELD, SD40:GREEN, SD40:PARK    SD1
13:   4   SD32:TRUMPET, SD1:FIELD, SD40:GREEN, SD40:PARK   SD40
14:   5               SD345:HOLE, SD340:DUST, SD387:ROLL  SD345
15:   5               SD345:HOLE, SD340:DUST, SD387:ROLL  SD340
16:   5               SD345:HOLE, SD340:DUST, SD387:ROLL  SD387
17:   6               SD42:CRAYON, SD345:FOX, SD183:WIRE   SD42
18:   6               SD42:CRAYON, SD345:FOX, SD183:WIRE  SD345
19:   6               SD42:CRAYON, SD345:FOX, SD183:WIRE  SD183
20:   7 SD455:TOMATO, SD86:RAY, SD39:MATURE, SD86:COSMIC  SD455
21:   7 SD455:TOMATO, SD86:RAY, SD39:MATURE, SD86:COSMIC   SD86
22:   7 SD455:TOMATO, SD86:RAY, SD39:MATURE, SD86:COSMIC   SD39

I would also like to split the elements in sdw.我还想拆分 sdw 中的元素。 But when I am trying that, it all gets messed up due to the fact that number of elements in sd and sdw are not necessarily the same.但是当我尝试这样做时，由于sd和sdw中的元素数量不一定相同，这一切都变得一团糟。

t <- G[ , list(sd = unlist(strsplit(sd, "," )),
               sdw = unlist(strsplit(sdw, "," ))) , by = list(GNO) ]

It give the warning它发出警告

Warning message:
In `[.data.table`(G, , list(sd = unlist(strsplit(sd, ",")), sdw = unlist(strsplit(sdw,  :
  Column 1 of result for group 4 is length 3 but the longest column in this result is 4. Recycled leaving remainder of 1 items. This warning is once only for the first group with this issue.

The output I am looking for is as follows.我正在寻找的输出如下。

out <- structure(list(GNO = c(1, 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 
5, 5, 5, 6, 6, 6, 7, 7, 7), sd = structure(c(1L, 14L, 9L, 6L, 
2L, 5L, 10L, 4L, 1L, 16L, 6L, 1L, 12L, 8L, 7L, 10L, 13L, 8L, 
3L, 15L, 17L, 11L), .Label = c("SD1", "SD12", "SD183", "SD23", 
"SD315", "SD32", "SD340", "SD345", "SD384", "SD387", "SD39", 
"SD40", "SD42", "SD44", "SD455", "SD567", "SD86"), class = "factor"), 
    sdw = structure(c(2L, 17L, 11L, 7L, 3L, 6L, 13L, 5L, 2L, 
    19L, 7L, 1L, 15L, 10L, 8L, 12L, 16L, 9L, 4L, 18L, 20L, 14L
    ), .Label = c("SD1:FIELD", "SD1:LUSH", "SD12:PAINTING", "SD183:WIRE", 
    "SD23:SWITCH", "SD315:MONEY31", "SD32:TRUMPET", "SD340:DUST", 
    "SD345:FOX", "SD345:HOLE", "SD384:FR563", "SD387:ROLL", "SD387:SPRING", 
    "SD39:MATURE", "SD40:GREEN, SD40:PARK", "SD42:CRAYON", "SD44:CANCEL", 
    "SD455:TOMATO", "SD567:TREK", "SD86:COSMIC, SD86:RAY"), class = "factor")), .Names = c("GNO", 
"sd", "sdw"), row.names = c(NA, -22L), class = "data.frame")

out
   GNO    sd                   sdw
1    1   SD1              SD1:LUSH
2    1  SD44           SD44:CANCEL
3    1 SD384           SD384:FR563
4    1  SD32          SD32:TRUMPET
5    2  SD12         SD12:PAINTING
6    2 SD315         SD315:MONEY31
7    2 SD387          SD387:SPRING
8    3  SD23           SD23:SWITCH
9    3   SD1              SD1:LUSH
10   3 SD567            SD567:TREK
11   4  SD32          SD32:TRUMPET
12   4   SD1             SD1:FIELD
13   4  SD40 SD40:GREEN, SD40:PARK
14   5 SD345            SD345:HOLE
15   5 SD340            SD340:DUST
16   5 SD387            SD387:ROLL
17   6  SD42           SD42:CRAYON
18   6 SD345             SD345:FOX
19   6 SD183            SD183:WIRE
20   7 SD455          SD455:TOMATO
21   7  SD86 SD86:COSMIC, SD86:RAY
22   7  SD39           SD39:MATURE

How to get this output?如何获得这个输出？

2 个解决方案

解决方案1
 6 已采纳  2014-12-31 12:46:58

Here's how I would approach this using only data.table package (no need in creating G2 neither)这是我将如何仅使用data.table包来解决这个问题（也不需要创建G2 ）

Here we are first splitting by swd and then aggregating by both GNO and the part before : in the stw column在这里，我们首先按swd拆分，然后按GNO和之前的部分聚合:在stw列中

G[, .(unlist(strsplit(sdw, ", ", TRUE))), GNO][, 
    .(stw = toString(V1)), .(GNO, sd = gsub(":.*", "", V1))]

 #    GNO    sd                   stw
 # 1:   1   SD1              SD1:LUSH
 # 2:   1  SD44           SD44:CANCEL
 # 3:   1 SD384           SD384:FR563
 # 4:   1  SD32          SD32:TRUMPET
 # 5:   2  SD12         SD12:PAINTING
 # 6:   2 SD315         SD315:MONEY31
 # 7:   2 SD387          SD387:SPRING
 # 8:   3  SD23           SD23:SWITCH
 # 9:   3   SD1              SD1:LUSH
# 10:   3 SD567            SD567:TREK
# 11:   4  SD32          SD32:TRUMPET
# 12:   4   SD1             SD1:FIELD
# 13:   4  SD40 SD40:GREEN, SD40:PARK
# 14:   5 SD345            SD345:HOLE
# 15:   5 SD340            SD340:DUST
# 16:   5 SD387            SD387:ROLL
# 17:   6  SD42           SD42:CRAYON
# 18:   6 SD345             SD345:FOX
# 19:   6 SD183            SD183:WIRE
# 20:   7 SD455          SD455:TOMATO
# 21:   7  SD86 SD86:RAY, SD86:COSMIC
# 22:   7  SD39           SD39:MATURE

解决方案2
 2  2014-12-31 12:23:15

This may be making too big of an assumption about the nature of your data, but it appears that the "sd" column is actually somewhat irrelevant since that information is already embedded in the "sdw" column.这可能对您的数据性质做出了过大的假设，但似乎“sd”列实际上有些无关紧要，因为该信息已嵌入“sdw”列中。

As such, this is what I came up with, using two cSplit s from my "splitstackshape" package:因此，这就是我想出的，使用我的“splitstackshape”包中的两个cSplit ：

library(splitstackshape)
temp <- cSplit(
  cSplit(
    G, "sdw", ",", "long"), 
  "sdw", ":")[, c(1, 3, 4)
              ][, list(paste(unique(sdw_1), unique(sdw_2), 
                             sep = ":", collapse = ", ")), 
                by = list(GNO, sdw_1)]
setnames(temp, c("GNO", "sd", "sdw"))
temp
#     GNO    sd                   sdw
#  1:   1   SD1              SD1:LUSH
#  2:   1  SD44           SD44:CANCEL
#  3:   1 SD384           SD384:FR563
#  4:   1  SD32          SD32:TRUMPET
#  5:   2  SD12         SD12:PAINTING
#  6:   2 SD315         SD315:MONEY31
#  7:   2 SD387          SD387:SPRING
#  8:   3  SD23           SD23:SWITCH
#  9:   3   SD1              SD1:LUSH
# 10:   3 SD567            SD567:TREK
# 11:   4  SD32          SD32:TRUMPET
# 12:   4   SD1             SD1:FIELD
# 13:   4  SD40 SD40:GREEN, SD40:PARK
# 14:   5 SD345            SD345:HOLE
# 15:   5 SD340            SD340:DUST
# 16:   5 SD387            SD387:ROLL
# 17:   6  SD42           SD42:CRAYON
# 18:   6 SD345             SD345:FOX
# 19:   6 SD183            SD183:WIRE
# 20:   7 SD455          SD455:TOMATO
# 21:   7  SD86 SD86:RAY, SD86:COSMIC
# 22:   7  SD39           SD39:MATURE
#     GNO    sd                   sdw

Comparison with your desired output与您想要的输出进行比较

library(compare)
compare(out, temp, allowAll = TRUE)
# TRUE 
#   [GNO] coerced from <integer> to <numeric>
#   [sdw] coerced from <character> to <factor>
#   dropped attributes
#   [3] dropped attributes

暂无
暂无

声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

相关问题 在 data.table 列中拆分文本字符串 - Split text string in a data.table columns data.table：使用矩阵更新data.table中的多个列 - data.table: Update multiple columns in a data.table with a matrix 拆分data.table - split data.table 在R中向data.table添加多个列 - Adding multiple columns to data.table in R 动态地按多列过滤data.table - Filter data.table by multiple columns, dynamically 替换data.table中多列中的值 - Replacing value in multiple columns in data.table ddply到data.table中等效的多个列 - ddply to multiple columns equivalent in data.table 更新data.table中的多列 - Updating multiple columns in data.table 从data.table中删除多个列 - Remove multiple columns from data.table 基于多列的数据表排序 - Sorting Data.Table Based on Multiple Columns
相关标签
 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM