r - 用分组变量中存在的值替换 NA

Question

虚假数据

fruit <- c("Orange", "Banana", "Orange", "Banana")
flavour <- c("Bitter", NA, NA, "Sweet")
geo <- c(NA, NA, NA, "France")
value <- c(1, NA, NA, 4)
dd <- data.frame(fruit, flavour, geo, value)
rm(fruit, flavour, geo, value)

我想按“水果”对数据集进行分组并替换缺失值在具有分组数据中存在的值的所有变量中。

期望的输出

fruit <- c("Orange", "Banana", "Orange", "Banana")
flavour <- c("Bitter", "Sweet", "Bitter", "Sweet")
geo <- c(NA, "France", NA, "France")
value <- c(1, 4, 1, 4)
dd2 <- data.frame(fruit, flavour, geo, value)
rm(fruit, flavour, geo, value)

代码尝试

tt <- dd %>%
  group_by(fruit) %>%
  summarise_all()

Answer 1

group_by 之后我们需要mutate_all(summarise/summarise_all 用于从多行中获取单行)。使用 zoo 中的 na.locf 为每一列用相邻的非 NA 元素填充 NA 元素

library(zoo)
library(dplyr)
dd %>% 
  group_by(fruit) %>% 
  mutate_all(funs(na.locf(na.locf(., na.rm =  FALSE), 
                 fromLast = TRUE, na.rm = FALSE)))
# A tibble: 4 x 4
# Groups:   fruit [2]
#  fruit  flavour geo    value
#  <fct>  <fct>   <fct>  <dbl>
#1 Orange Bitter  <NA>       1
#2 Banana Sweet   France     4
#3 Orange Bitter  <NA>       1
#4 Banana Sweet   France     4