在R中按组合并数据

我构造以下data.frame对象：

name <- c("Homer", "Marge", "Bart", "Lisa", "Maggie")
incidents <- c(133, 36, 1242, 2, NA)
gender <- c("MALE", "FEMALE", "MALE", "FEMALE", "FEMALE")
data <- data.frame(name, incidents, gender)

首先，我使用

clean_data <- data[!is.na(incidents), ]

现在，我按性别与

agg <- aggregate(incidents ~ gender, clean_data, mean)

现在，我想用agg的数据填充事件中的NA值，这样数据=

    name incidents gender
1  Homer       133   MALE
2  Marge        36 FEMALE
3   Bart      1242   MALE
4   Lisa         2 FEMALE
5 Maggie      19.0 FEMALE

使用base R最简单的方法是什么？

您可以使用ave。它以与原始数据集中相同的顺序提供平均值VAL，检查事件列中的NA元素，并用相应NA元素的VAL替换这些元素

 vals <- with(data, ave(incidents, gender, FUN= function(x)
                                         mean(x, na.rm=TRUE)))
 indx1 <- is.na(data$incidents)
 data$incidents[indx1] <- vals[indx1]

如@MrFlick在评论中所示的较短版本。使用ifelse，它将NA元素替换为平均值

 data$incidents<-with(data, ave(incidents, gender,
          FUN=function(x) ifelse(is.na(x), mean(x, na.rm=T), x)))

---

代替ifelse，replace也可以用data.table显示为@Ananda Mahto

对于多样性，这里有一种使用data.table的方法，它还演示了replace函数

library(data.table)
as.data.table(data)[
  , incidents := replace(incidents, is.na(incidents), 
                         mean(incidents, na.rm = TRUE)), 
  by = gender][]
#      name incidents gender
# 1:  Homer       133   MALE
# 2:  Marge        36 FEMALE
# 3:   Bart      1242   MALE
# 4:   Lisa         2 FEMALE
# 5: Maggie        19 FEMALE

---

我也有同样的想法，但没有数据$incidents@MrFlick看起来好多了。您可以将其作为新答案发布。事实上，我在做ave之前没有看过结果。老实说，我认为在完整的data.frame上使用ave是这里的秘密。我的回答显得多余。那么完整的解决方案应该是什么呢？我是否执行了太多步骤？@GeoffLittle MrFlick的版本是最短的。

 vals <- with(data, ave(incidents, gender, FUN= function(x)
                                         mean(x, na.rm=TRUE)))
 indx1 <- is.na(data$incidents)
 data$incidents[indx1] <- vals[indx1]

 data$incidents<-with(data, ave(incidents, gender,
          FUN=function(x) ifelse(is.na(x), mean(x, na.rm=T), x)))

library(data.table)
as.data.table(data)[
  , incidents := replace(incidents, is.na(incidents), 
                         mean(incidents, na.rm = TRUE)), 
  by = gender][]
#      name incidents gender
# 1:  Homer       133   MALE
# 2:  Marge        36 FEMALE
# 3:   Bart      1242   MALE
# 4:   Lisa         2 FEMALE
# 5: Maggie        19 FEMALE