Question

我观察不同日期（日期2）的人口中不同角色（代码）的频率（eff）。

datas <- data.frame(date2 = rep(seq(Sys.Date() - 2, Sys.Date(), by = "day"), each = 2), 
                    date1 = Sys.Date(), 
                    code = rep(LETTERS[1:2], 3), 
                    eff = c(50, 30, 20, 10, 20, 20), 
                    total = 100)

> datas
       date2      date1 code eff total
1 2015-07-25 2015-07-27    A  50   100
2 2015-07-25 2015-07-27    B  30   100
3 2015-07-26 2015-07-27    A  20   100
4 2015-07-26 2015-07-27    B  10   100
5 2015-07-27 2015-07-27    A  20   100
6 2015-07-27 2015-07-27    B  20   100

对于每个日期2我想添加另一个代码，“填写”今天的频率总和与当天观察到的总人口（总数）之间的差异。

例如，这是我期望的输出：

       date2 eff      date1 total code
1 2015-07-25  50 2015-07-27   100    A
2 2015-07-25  30 2015-07-27   100    B
3 2015-07-25  20 2015-07-27   100   KO
4 2015-07-26  20 2015-07-27   100    A
5 2015-07-26  10 2015-07-27   100    B
6 2015-07-26  70 2015-07-27   100   KO
7 2015-07-27  20 2015-07-27   100    A
8 2015-07-27  20 2015-07-27   100    B
9 2015-07-27  60 2015-07-27   100   KO

这就是我产生输出的方式：

datas %>% 
  group_by(date2) %>% 
  summarise(eff = (sum(total) / n()) - sum(eff)) %>% 
  inner_join(datas, by = "date2") %>% 
  select(-c(eff.y, code), eff = eff.x) %>% 
  distinct %>% 
  mutate(code = "KO") %>% 
  bind_rows(datas)

但我不喜欢这个解决方案，我想知道是否有人有更好的解决方案！

另外，如何处理2个分组变量（例如下面的例子中的date1＆amp; date 2）？

datas2 <- data.frame(date2 = c(rep(seq(Sys.Date() - 2, Sys.Date(), by = "day"), each = 2), 
                              rep(seq(Sys.Date() - 1, Sys.Date(), by = "day"), each = 2)), 
                    date1 = c(rep(Sys.Date() - 3, 6), rep(Sys.Date() - 2, 4)), 
                    code = c(rep(LETTERS[1:2], 3), rep(LETTERS[1:2], 2)), 
                    eff = c(50, 30, 20, 10, 20, 20, 10, 20, 30, 40), 
                    total = 100)
> datas2
        date2      date1 code eff total
1  2015-07-25 2015-07-24    A  50   100
2  2015-07-25 2015-07-24    B  30   100
3  2015-07-26 2015-07-24    A  20   100
4  2015-07-26 2015-07-24    B  10   100
5  2015-07-27 2015-07-24    A  20   100
6  2015-07-27 2015-07-24    B  20   100
7  2015-07-26 2015-07-25    A  10   100
8  2015-07-26 2015-07-25    B  20   100
9  2015-07-27 2015-07-25    A  30   100
10 2015-07-27 2015-07-25    B  40   100

感谢您的任何想法！

Answer 1

您可以尝试

 library(dplyr)
 datas %>% 
   group_by(date2, date1) %>% 
   summarise(eff=total[1L]-sum(eff), code='KO', total=total[1L]) %>%
   bind_rows(., datas) %>%
   arrange(date2,code)

或类似的data.table方法

 library(data.table)
 rbind(datas,setDT(datas)[, list(eff=total[1L]-sum(eff),
           code='KO',total=total[1L]),.(date2,date1)])[order(date2)]

按组添加行

1 个答案: