我想在dplyr中使用c()作为聚合函数来聚合一些字符串。我首先尝试了以下内容:
> InsectSprays$spray = as.character(InsectSprays$spray)
> dt = tbl_df(InsectSprays)
> dt %>% group_by(count) %>% summarize(c(spray))
Error: expecting a single value
但是在aggregate()中使用c()函数可以工作:
> da = aggregate(spray ~ count, InsectSprays, c)
> head(da)
count spray
1 0 C, C
2 1 C, C, C, C, E, E
3 2 C, C, D, E>
在stackoverflow hinted中搜索而不是c()函数,使用带崩溃的paste()可以解决问题:
dt %>% group_by(count) %>% summarize(s=paste(spray, collapse=","))
或
dt %>% group_by(count) %>% summarize(paste( c(spray), collapse=","))
我的问题是:为什么c()函数在aggregate()中工作但在dplyr summary()中不起作用?
答案 0 :(得分:5)
如果仔细观察,我们会发现c()
在我们使用do()
时确实有效(在某种程度上)。但据我了解,dplyr
目前不允许这种类型的列表打印
> InsectSprays$spray = as.character(InsectSprays$spray)
> dt = tbl_df(InsectSprays)
> doC <- dt %>% group_by(count) %>% do(s = c(.$spray))
> head(doC)
Source: local data frame [6 x 2]
count s
1 0 <chr[2]>
2 1 <chr[6]>
3 2 <chr[4]>
4 3 <chr[8]>
5 4 <chr[4]>
6 5 <chr[7]>
> head(doC)[[2]]
[[1]]
[1] "C" "C"
[[2]]
[1] "C" "C" "C" "C" "E" "E"
[[3]]
[1] "C" "C" "D" "E"
[[4]]
[1] "C" "C" "D" "D" "E" "E" "E" "E"
[[5]]
[1] "C" "D" "D" "E"
[[6]]
[1] "D" "D" "D" "D" "D" "E" "E"