Question

我有一个数据框df（请参阅下面的代码），其中包含近100,000行，显示了我的程序联系人列表。该列表包含一个列，其中显示与联系人关联的程序program和组织O_ID以及联系人在程序中具有的角色列。每当联系人在多个程序中或在程序中具有多个角色时，将为该联系人创建另一行，其中程序和联系人角色字段值会发生变化。

First   Last    C_ID    OrgName O_ID Program    Role
John    Smith   10045   Acme    901 Buildings   Primary
John    Smith   10045   Acme    901 Buildings   Communications
John    Smith   10045   Acme    901 Homes       Primary
Teddy   Bush    10046   Acme    901 Buildings   Primary
Teddy   Bush    10046   Acme    901 Buildings   Signatory
Jess    Clinton 10050   Consult 904 Homes       Signatory
Jess    Clinton 10050   Consult 904 Homes       Primary
Jess    Clinton 10050   Consult 904 Homes       Communications

出于演示目的，我试图尽量减少行数。具体来说，如果联系人在同一个组织和同一个程序中，我只希望联系人出现在一行上（此时为几行），并将联系人角色组合成一个字符串。

我尝试了这段代码并且部分有效：ddply(df,.(df$C_ID, df$Program, df$O_ID), paste, sep=",")

结果如下：

df$C_ID df$Program df$O_ID                        V1                                 V2
1       10045      Buildings         901         c("John", "John")                c("Smith", "Smith")
2       10045          Homes         901                      John                              Smith
3       10046      Buildings         901       c("Teddy", "Teddy")                  c("Bush", "Bush")
4       10050          Homes         904 c("Jess", "Jess", "Jess") c("Clinton", "Clinton", "Clinton")
                      V3                                 V4               V5                           V6
1        c(10045, 10045)                  c("Acme", "Acme")      c(901, 901)  c("Buildings", "Buildings")
2                  10045                               Acme              901                        Homes
3        c(10046, 10046)                  c("Acme", "Acme")      c(901, 901)  c("Buildings", "Buildings")
4 c(10050, 10050, 10050) c("Consult", "Consult", "Consult") c(904, 904, 904) c("Homes", "Homes", "Homes")
                                           V7
1              c("Primary", "Communications")
2                                     Primary
3                   c("Primary", "Signatory")
4 c("Signatory", "Primary", "Communications")

问题是

1）重新排列了列（请注意我的实际数据集中有更多列），列名称消失了

2）唯一具有更改值的列应位于Role列中。但是，即使合并的值相同，结果也会将大多数列的值组合在一起。例如，在结果列V1（第一个名称列）中，返回c("John", "John")。它应该只读“约翰”。应该具有不同值的唯一列是列V7 c("Primary", "Communications")

df<-structure(list(First = c("John", "John", "John", "Teddy", "Teddy", 
"Jess", "Jess", "Jess"), Last = c("Smith", "Smith", "Smith", 
"Bush", "Bush", "Clinton", "Clinton", "Clinton"), C_ID = c(10045L, 
10045L, 10045L, 10046L, 10046L, 10050L, 10050L, 10050L), OrgName = c("Acme", 
"Acme", "Acme", "Acme", "Acme", "Consult", "Consult", "Consult"
), O_ID = c(901L, 901L, 901L, 901L, 901L, 904L, 904L, 904L), 
    Program = c("Buildings", "Buildings", "Homes", "Buildings", 
    "Buildings", "Homes", "Homes", "Homes"), Role = c("Primary", 
    "Communications", "Primary", "Primary", "Signatory", "Signatory", 
    "Primary", "Communications")), .Names = c("First", "Last", 
"C_ID", "OrgName", "O_ID", "Program", "Role"), class = "data.frame", row.names = c(NA, 
-8L))

Answer 1

paste中您需要的是collapse = ", "，而不是sep。使用collapse从所有输入创建单个字符串。我这样做是通过对所有标识列 - 名称，组织，程序等进行分组 - 然后在summarise中折叠角色。

library(tidyverse)

df %>%
  group_by(First, Last, C_ID, OrgName, O_ID, Program) %>%
  summarise(roles_mult = paste(Role, collapse = ", "))
#> # A tibble: 4 x 7
#> # Groups:   First, Last, C_ID, OrgName, O_ID [?]
#>   First Last     C_ID OrgName  O_ID Program   roles_mult                  
#>   <chr> <chr>   <int> <chr>   <int> <chr>     <chr>                       
#> 1 Jess  Clinton 10050 Consult   904 Homes     Signatory, Primary, Communi…
#> 2 John  Smith   10045 Acme      901 Buildings Primary, Communications     
#> 3 John  Smith   10045 Acme      901 Homes     Primary                     
#> 4 Teddy Bush    10046 Acme      901 Buildings Primary, Signatory

Answer 2

您也可以使用dplyr执行此操作。

> df %>% distinct(First, Last, .keep_all=T)
  First    Last  C_ID OrgName O_ID   Program      Role
1  John   Smith 10045    Acme  901 Buildings   Primary
2 Teddy    Bush 10046    Acme  901 Buildings   Primary
3  Jess Clinton 10050 Consult  904     Homes Signatory

根据条件连接行中的值

2 个答案: