R组或聚合

Question

我想做group_by或聚合。我有类似的东西：

> head(affiliation_clean)
  Affiliation_ID                       Affiliation_Name          City       Country
1       000001   New Mexico State University Las Cruces    Las Cruces United States
2       000001   New Mexico State University Las Cruces    Las Cruces          <NA>
3       000001   New Mexico State University Las Cruces          <NA>          <NA>
4       000002   Palo Alto Research Center Incorporated     Palo Alto          <NA>
5       000002   Palo Alto Research Center Incorporated          <NA> United States
6       000002   Palo Alto Research Center Incorporated          <NA>          <NA>

按“Affiliation_ID”分组并取最长的“Affiliation_Name”，“City”和“Country”字符串，我想得到：

> head(affiliation_clean)
  Affiliation_ID                       Affiliation_Name          City       Country
1       000001   New Mexico State University Las Cruces    Las Cruces United States
2       000002   Palo Alto Research Center Incorporated     Palo Alto United States

提前致谢。

Answer 1

假设每个'Affiliation_ID'，'Affiliation_Name'都有一个unique'城市/国家'，在前两列分组后，获取所有unique非NA元素summarise_all

的其他列

library(dplyr)
affiliation_clean %>%
   group_by(Affiliation_ID, Affiliation_Name) %>%
   summarise_all(funs(unique(.[!is.na(.)])) )
# A tibble: 2 x 4
# Groups: Affiliation_ID [?]
#  Affiliation_ID Affiliation_Name                       City       Country      
#  <chr>          <chr>                                  <chr>      <chr>        
#1 000001         New Mexico State University Las Cruces Las Cruces United States
#2 000002         Palo Alto Research Center Incorporated Palo Alto  United States

Answer 2

以下是基于您的说明的dplyr解决方案，用于选择每个Affiliation_ID和列的最长字符串。

library(dplyr)

dat2 <- dat %>%
  group_by(Affiliation_ID) %>%
  summarise_all(funs(.[which.max(nchar(.))][1]))
dat2
# # A tibble: 2 x 4
#   Affiliation_ID Affiliation_Name                       City       Country      
#            <int> <chr>                                  <chr>      <chr>        
# 1              1 New Mexico State University Las Cruces Las Cruces United States
# 2              2 Palo Alto Research Center Incorporated Palo Alto  United States

数据

dat <-read.table(text = "  Affiliation_ID                       Affiliation_Name          City       Country
1       '000001'   'New Mexico State University Las Cruces'    'Las Cruces' 'United States'
                 2       '000001'   'New Mexico State University Las Cruces'    'Las Cruces'          NA
                 3       '000001'   'New Mexico State University Las Cruces'          NA          NA
                 4       '000002'   'Palo Alto Research Center Incorporated'     'Palo Alto'          NA
                 5       '000002'   'Palo Alto Research Center Incorporated'          NA 'United States'
                 6       '000002'   'Palo Alto Research Center Incorporated'          NA          NA",
                 header = TRUE, stringsAsFactors = FALSE)