我希望val
中所有ctry
的{{1}}中的cat1
以特定ctry
方式归咎于遗漏。
数据示例
set.seed(654)
df1 <- data.frame(
year=rep(2000:2005, each=5),
ctry=rep(LETTERS[1:5], 6),
val=rnorm(30)
)
df1$cat <- ifelse(df1$ctry %in% c("A", "B"), 1, 0)
df1[sample(nrow(df1), 12), "val"] <- NA
> head(df1)
year ctry val cat
1 2000 A -0.76031762 1
2 2000 B -0.38970450 1
3 2000 C 1.68962523 0
4 2000 D NA 0
5 2000 E 0.09530146 0
6 2001 A NA 1
首先,我在ctry
中获取cat1
的名称并分配他们的资金。
cat1 <- as.character(sort(unique(
df1[!is.na(df1$val) & df1$cat == 1, ]
[, 2])))
cat1 <- sapply(cat1, function(x) mean(df1$val[df1$ctry == x], na.rm=TRUE))
> cat1
A B
0.4372003 0.4792314
现在我成功地按国家/地区手动输入:
df2 <- df1
df2$val[df2$ctry %in% names(cat1)[1] & is.na(df2$val)] <- cat1[1]
> head(df2)
year ctry val cat
1 2000 A -0.76031762 1
2 2000 B -0.38970450 1
3 2000 C 1.68962523 0
4 2000 D NA 0
5 2000 E 0.09530146 0
6 2001 A -0.49758245 1
但由于某种原因,我无法使sapply()
工作,自动进行估算:
> sapply(seq_along(cat1),
+ function(x) df2$val[df2$ctry %in% names(cat1)[x] & is.na(df2$val)] <- cat1[x])
A B
-0.4975825 -0.6139364
预期输出将是一个完整的数据框架,具有cat1
类别国家的特定估算方式。
答案 0 :(得分:1)
您可以使用tidyverse
尝试group_by
方法,以获得每ctry
的平均值。然后使用NA
更新ifelse
。添加了一个新列val2
来说明正在发生的事情。您可以编写"val"
来覆盖该列。
library(tidyverse)
df1 %>%
group_by(ctry) %>%
mutate(Mean=mean(val, na.rm = T)) %>%
mutate(val2=ifelse(is.na(val) & cat == 1, Mean, val)) %>%
ungroup()
# A tibble: 30 x 6
year ctry val cat Mean val2
<int> <fct> <dbl> <dbl> <dbl> <dbl>
1 2000 A -0.760 1 -0.498 -0.760
2 2000 B -0.390 1 -0.614 -0.390
3 2000 C 1.69 0 0.397 1.69
4 2000 D NA 0 -0.0321 NA
5 2000 E 0.0953 0 -0.513 0.0953
6 2001 A NA 1 -0.498 -0.498
7 2001 B NA 1 -0.614 -0.614
8 2001 C NA 0 0.397 NA
9 2001 D NA 0 -0.0321 NA
10 2001 E NA 0 -0.513 NA
# ... with 20 more rows
答案 1 :(得分:1)
如果我的理解是正确的,你想要自动完成最后一个过程
sapply(seq_along(cat1),
function(x) df2$val[df2$ctry %in% names(cat1)[x] & is.na(df2$val)] <<- cat1[x])
> df2
year ctry val cat
1 2000 A -0.760317618 1
2 2000 B -0.389704501 1
3 2000 C 1.689625228 0
4 2000 D NA 0
5 2000 E 0.095301460 0
6 2001 A -0.497582454 1
7 2001 B -0.613936417 1
8 2001 C NA 0
9 2001 D NA 0
10 2001 E NA 0
11 2002 A -0.107260116 1
12 2002 B -0.838168333 1
13 2002 C -0.982605890 0
14 2002 D -0.820370986 0
15 2002 E -0.871432562 0
16 2003 A -0.497582454 1
17 2003 B -0.613936417 1
18 2003 C -0.006557849 0
19 2003 D 0.661696295 0
20 2003 E -0.762828067 0
21 2004 A -0.286692466 1
22 2004 B -0.613936417 1
23 2004 C 0.512579937 0
24 2004 D 0.722127317 0
25 2004 E NA 0
26 2005 A -0.836059616 1
27 2005 B -0.613936417 1
28 2005 C 0.774016151 0
29 2005 D -0.691866605 0
30 2005 E NA 0
我只需将<
替换为<<-
scoping assignment
答案 2 :(得分:1)
在基地R:
set.seed(654)
df1 <- data.frame(
year=rep(2000:2005, each=5),
ctry=rep(LETTERS[1:5], 6),
val=rnorm(30)
)
df1$cat <- ifelse(df1$ctry %in% c("A", "B"), 1, 0)
df1[sample(nrow(df1), 12), "val"] <- NA
# want:
my_means <- tapply(df1$val, df1$ctry, mean, na.rm = TRUE)
df1$val <- ifelse(is.na(df1$val), my_means[df1$ctry], df1$val)