如何将存储在单个值（每个记录）中的字符列表转换为可以分析的格式？ R

Question

所以我有以下数据。在这个数据框中，我有一个用户名、日期和一个包列表。我的目标是将该列 packages 转换为我可以分析的格式。

> print(data_example)
# A tibble: 4 x 3
  username date     packages                                                                   
  <chr>    <chr>    <chr>                                                                      
1 John     1/5/2015 "compiler 4.1.2, magrittr 2.0.1, ellipsis 0.3.2, tools 4.1.2, pillar 1.6.4"
2 Karen    1/5/2015 "compiler 4.1.2, tools 4.1.2\""                                            
3 Mike     1/5/2015 "evaluate 0.14, highr 0.9, httr 1.4.2, pillar 1.6.4, rlang 0.4.12"         
4 Zoe      1/6/2015 "httr 1.4.2, viridisLite 0.4.0, jsonlite 1.7.2, splines 4.1.2"

我试过做这样的事情，但这并没有给我多大价值。

packages <- as.list(parsed_log$packages)

我的目标是尝试“延长支点”并获得这样的结果？我不确定这是否是分析此类数据的最佳方法。

username   date      packages
John       1/5/2015  compiler 4.1.2
John       1/5/2015  magrittr 2.0.1
John       1/5/2015  ellipsis 0.3.2
John       1/5/2015  tools 4.1.2
ect...

Answer 1

一个解决方案，基于tidyr::separate：

library(tidyr)

df <- data.frame(
  stringsAsFactors = FALSE,
  username = c("John", "Karen", "Mike", "Zoe"),
  date = c("1/5/2015", "1/5/2015", "1/5/2015", "1/6/2015"),
  packages = c("compiler 4.1.2, magrittr 2.0.1, ellipsis 0.3.2, tools 4.1.2, pillar 1.6.4","compiler 4.1.2, tools 4.1.2\\"",
               "evaluate 0.14, highr 0.9, httr 1.4.2, pillar 1.6.4, rlang 0.4.12",
               "httr 1.4.2, viridisLite 0.4.0, jsonlite 1.7.2, splines 4.1.2")
)

separate_rows(df, packages, sep = ", ")

#> # A tibble: 16 × 3
#>    username date     packages           
#>    <chr>    <chr>    <chr>              
#>  1 John     1/5/2015 "compiler 4.1.2"   
#>  2 John     1/5/2015 "magrittr 2.0.1"   
#>  3 John     1/5/2015 "ellipsis 0.3.2"   
#>  4 John     1/5/2015 "tools 4.1.2"      
#>  5 John     1/5/2015 "pillar 1.6.4"     
#>  6 Karen    1/5/2015 "compiler 4.1.2"   
#>  7 Karen    1/5/2015 "tools 4.1.2\\""  
#>  8 Mike     1/5/2015 "evaluate 0.14"    
#>  9 Mike     1/5/2015 "highr 0.9"        
#> 10 Mike     1/5/2015 "httr 1.4.2"       
#> 11 Mike     1/5/2015 "pillar 1.6.4"     
#> 12 Mike     1/5/2015 "rlang 0.4.12"     
#> 13 Zoe      1/6/2015 "httr 1.4.2"       
#> 14 Zoe      1/6/2015 "viridisLite 0.4.0"
#> 15 Zoe      1/6/2015 "jsonlite 1.7.2"   
#> 16 Zoe      1/6/2015 "splines 4.1.2"

Answer 2

一种选择是使用stringi拆分：

s <- strsplit(as.character(df$packages), ',')
data.frame(
  packages = unlist(s),
  date = rep(df$date, length(s)),
  username = rep(df$username, length(s))
)

             packages     date username
1      compiler 4.1.2 1/5/2015     John
2      magrittr 2.0.1 1/5/2015    Karen
3      ellipsis 0.3.2 1/5/2015     Mike
4         tools 4.1.2 1/6/2015      Zoe
5        pillar 1.6.4 1/5/2015     John
6      compiler 4.1.2 1/5/2015    Karen
7      tools 4.1.2\" 1/5/2015     Mike
8       evaluate 0.14 1/6/2015      Zoe
9           highr 0.9 1/5/2015     John
10         httr 1.4.2 1/5/2015    Karen
11       pillar 1.6.4 1/5/2015     Mike
12       rlang 0.4.12 1/6/2015      Zoe
13         httr 1.4.2 1/5/2015     John
14  viridisLite 0.4.0 1/5/2015    Karen
15     jsonlite 1.7.2 1/5/2015     Mike
16      splines 4.1.2 1/6/2015      Zoe

另一种选择是使用 cSplit 来自 splitstackshape:

library(splitstackshape)
cSplit(df, "packages", ",", direction = "long")

data.table 选项：

library(data.table)
dt <- as.data.table(df)
dt[, strsplit(as.character(packages), ","), by = c("username", "date")]

数据（感谢@PaulS 提供数据）

df <- structure(list(username = c("John", "Karen", "Mike", "Zoe"), 
                     date = c("1/5/2015", "1/5/2015", "1/5/2015", "1/6/2015"), 
                     packages = c("compiler 4.1.2, magrittr 2.0.1, ellipsis 0.3.2, tools 4.1.2, pillar 1.6.4", 
                                  "compiler 4.1.2, tools 4.1.2\\"", "evaluate 0.14, highr 0.9, httr 1.4.2, pillar 1.6.4, rlang 0.4.12", 
                                  "httr 1.4.2, viridisLite 0.4.0, jsonlite 1.7.2, splines 4.1.2"
                     )), class = "data.frame", row.names = c(NA, -4L))

如何将存储在单个值（每个记录）中的字符列表转换为可以分析的格式？ R

How to convert a list of characters that is stored in a single value (per record) into a format that can be analyzed? R

r

dplyr

tidyr