将 NA 转换为基于另一列的最常出现的值

Question

我有一个名为 df 的数据框，如下所示：

  Author_ID Country Cited Name  Title
1: 1        Spain   10    Alex  Whatever
2: 1        France  15    Ale   Whatever2
3: 1        NA      10    Alex  Whatever3
4: 1        Spain   10    Alex  Whatever4
5: 2        Italy   10    Alice Whatever5
6: 2        Greece  10    Alice Whatever6
7: 2        Greece  10    Alice Whatever7
8: 2        NA      10    Alce  Whatever8
8: 2        NA      10    Alce  Whatever8

我想得到这样的东西，其中 NA 被替换为那个 Author_ID 出现次数最多的国家（如果有两个国家出现相同次数，随机那两个会很好）：

Author_ID Country Cited Name  Title
    1: 1        Spain   10    Alex  Whatever
    2: 1        France  15    Ale   Whatever2
    3: 1        Spain   10    Alex  Whatever3
    4: 1        Spain   10    Alex  Whatever4
    5: 2        Italy   10    Alice Whatever5
    6: 2        Greece  10    Alice Whatever6
    7: 2        Greece  10    Alice Whatever7
    8: 2        Greece  10    Alce  Whatever8
    8: 2        Greece  10    Alce  Whatever8

提前致谢。

Answer 1

与data.table

library(data.table)
# setDT(df)
df[,Country := replace(Country,is.na(Country),names(which.max(table(Country)))),by=Author_ID]

#    Author_ID Country Cited  Name     Title
# 1:         1   Spain    10  Alex  Whatever
# 2:         1  France    15   Ale Whatever2
# 3:         1   Spain    10  Alex Whatever3
# 4:         1   Spain    10  Alex Whatever4
# 5:         2   Italy    10 Alice Whatever5
# 6:         2  Greece    10 Alice Whatever6
# 7:         2  Greece    10 Alice Whatever7
# 8:         2  Greece    10  Alce Whatever8
# 9:         2  Greece    10  Alce Whatever8

在基地 R:

df$Country <- unlist(tapply(df$Country,df$Author_ID,function(x)
  replace(x,is.na(x),names(which.max(table(x))))))
# Author_ID Country Cited  Name     Title
# 1         1   Spain    10  Alex  Whatever
# 2         1  France    15   Ale Whatever2
# 3         1   Spain    10  Alex Whatever3
# 4         1   Spain    10  Alex Whatever4
# 5         2   Italy    10 Alice Whatever5
# 6         2  Greece    10 Alice Whatever6
# 7         2  Greece    10 Alice Whatever7
# 8         2  Greece    10  Alce Whatever8
# 9         2  Greece    10  Alce Whatever8

与 dplyr:

library(dplyr)
df %>% group_by(Author_ID) %>%
  mutate(Country = replace(
    Country,
    is.na(Country),
    names(which.max(table(Country)))))

# # A tibble: 9 x 5
# # Groups:   Author_ID [2]
# Author_ID Country Cited  Name     Title
# <int>   <chr> <int> <chr>     <chr>
# 1         1   Spain    10  Alex  Whatever
# 2         1  France    15   Ale Whatever2
# 3         1   Spain    10  Alex Whatever3
# 4         1   Spain    10  Alex Whatever4
# 5         2   Italy    10 Alice Whatever5
# 6         2  Greece    10 Alice Whatever6
# 7         2  Greece    10 Alice Whatever7
# 8         2  Greece    10  Alce Whatever8
# 9         2  Greece    10  Alce Whatever8

如果多个国家出现的时间达到最大值，则取第一个，而不是随机的。

如果某些国家/地区仅适用于某些作者

首先调用它来修改示例数据：

df$Country[df$Author_ID ==2] <- NA

那么这里是 3 个改编的解决方案，不是非常优雅，但它有效。我怀疑可能有一个 base/dplyr/data.table 函数可以更顺利地将零长度元素更改为 NA。

setDT(df)
df[,Country := replace(Country,is.na(Country),{
  nm <- names(which.max(table(x)))
  if(length(nm)==0) NA else nm}),
  by=Author_ID]
df <- df[!is.na(df$Country),]

#    Author_ID Country Cited Name     Title
# 1:         1   Spain    10 Alex  Whatever
# 2:         1  France    15  Ale Whatever2
# 3:         1   Spain    10 Alex Whatever4

df$Country <- unlist(tapply(df$Country,df$Author_ID,function(x)
  replace(x,is.na(x),{
    nm <- names(which.max(table(x)))
    if(length(nm)==0) NA else nm
    })))
df <- df[!is.na(df$Country),]

# Author_ID Country Cited Name     Title
# 1         1   Spain    10 Alex  Whatever
# 2         1  France    15  Ale Whatever2
# 3         1   Spain    10 Alex Whatever3
# 4         1   Spain    10 Alex Whatever4

df %>% group_by(Author_ID) %>%
  mutate(Country = replace(
    Country,
    is.na(Country),
    names(which.max(table(Country))) %>%
      {if(length(.)==0) NA else .})) %>%
  filter(!is.na(Country))

# # A tibble: 4 x 5
# # Groups:   Author_ID [1]
# Author_ID Country Cited Name  Title    
# <int> <chr>   <int> <chr> <chr>    
# 1         1 Spain      10 Alex  Whatever 
# 2         1 France     15 Ale   Whatever2
# 3         1 Spain      10 Alex  Whatever3
# 4         1 Spain      10 Alex  Whatever4

数据

df <- read.table(text="Author_ID Country Cited Name  Title
1        Spain   10    Alex  Whatever
1        France  15    Ale   Whatever2
1        NA      10    Alex  Whatever3
1        Spain   10    Alex  Whatever4
2        Italy   10    Alice Whatever5
2        Greece  10    Alice Whatever6
2        Greece  10    Alice Whatever7
2        NA      10    Alce  Whatever8
2        NA      10    Alce  Whatever8",h=T,strin=F)

将 NA 转换为基于另一列的最常出现的值

Convert NA to most appearing value based in another column

r

dataframe

na