通过使用数据表和 R 对新组进行分组和重置计数器来计算行数答案

【问题标题】：Count rows by grouping and reset counter on new group with datatable and R通过使用数据表和 R 对新组进行分组和重置计数器来计算行数
【发布时间】：2022-11-25 05:53:57
【问题描述】：

我需要计算数据表获取的每个数据池的行数。这些要求必须填写周列“week”和“Exist”。我有以下数据：

week_2020 <- seq(202001, 202015, 1)
week_2021 <- seq(202101, 202110, 1)

Exist <- c("TRUE","TRUE","TRUE","TRUE","TRUE","TRUE","TRUE","TRUE",
           NA,NA,NA,
           "TRUE","TRUE",NA,NA,"TRUE","TRUE","TRUE","TRUE",
           NA,NA,NA,
           "TRUE",NA, NA)

Year <- c(rep(2020,15),rep(2021,10) )

df<-data.table(Store = "store_1", Client = "client_1", 
               Year = Year,
               week = c(week_2020, week_2021),
               Exist = Exist)

Store	Client	Year	Week	Exist
Store 1	Client 1	2020	202001	TRUE
Store 1	Client 1	2020	202002	TRUE
Store 1	Client 1	2020	202003	TRUE
Store 1	Client 1	2020	202004	TRUE
Store 1	Client 1	2020	202005	TRUE
Store 1	Client 1	2020	202006	TRUE
Store 1	Client 1	2020	202007	TRUE
Store 1	Client 1	2020	202008	TRUE
Store 1	Client 1	2020	202009	NA
Store 1	Client 1	2020	202010	NA
Store 1	Client 1	2020	202011	NA
Store 1	Client 1	2020	202012	TRUE
Store 1	Client 1	2020	202013	TRUE
Store 1	Client 1	2020	202014	NA
Store 1	Client 1	2020	202015	NA
Store 1	Client 1	2021	202101	TRUE
Store 1	Client 1	2021	202102	TRUE
Store 1	Client 1	2021	202103	TRUE
Store 1	Client 1	2021	202104	TRUE
Store 1	Client 1	2021	202105	NA
Store 1	Client 1	2021	202106	NA
Store 1	Client 1	2021	202107	NA
Store 1	Client 1	2021	202108	TRUE
Store 1	Client 1	2021	202109	NA
Store 1	Client 1	2021	202110	NA

如表所示，“存在”一栏的部分数据为NA，表示不存在，但该分组也要算进去。

我创建了一个变量来帮助我计算周数，直到我找到丢失的周数，然后我应该计算它们并重置计数器，另一个变量计算最大计数“n_week_Count”，但我不能做我需要的。我希望你能帮我解决这个问题。谢谢指教

这就是我所拥有的...

df[, ':=' (n_weekCount = 1:.SD[,(.N)] ), keyby = c("Store", "Client", "Year", "Exist")
   ][, ':=' (MaxweekCount = .SD[, max(n_weekCount)]), keyby = c("Store", "Client", "Year", "Exist")
][order(week)]

Store	Client	Year	Week	Exist	n_weekCount	maxWeek_Count
Store 1	Client 1	2020	202001	TRUE	1	10
Store 1	Client 1	2020	202002	TRUE	2	10
Store 1	Client 1	2020	202003	TRUE	3	10
Store 1	Client 1	2020	202004	TRUE	4	10
Store 1	Client 1	2020	202005	TRUE	5	10
Store 1	Client 1	2020	202006	TRUE	6	10
Store 1	Client 1	2020	202007	TRUE	7	10
Store 1	Client 1	2020	202008	TRUE	8	10
Store 1	Client 1	2020	202009	NA	1	5
Store 1	Client 1	2020	202010	NA	2	5
Store 1	Client 1	2020	202011	NA	3	5
Store 1	Client 1	2020	202012	TRUE	9	10
Store 1	Client 1	2020	202013	TRUE	10	10
Store 1	Client 1	2020	202014	NA	4	5
Store 1	Client 1	2020	202015	NA	5	5
Store 1	Client 1	2021	202101	TRUE	1	10
Store 1	Client 1	2021	202102	TRUE	2	10
Store 1	Client 1	2021	202103	TRUE	3	10
Store 1	Client 1	2021	202104	TRUE	4	10
Store 1	Client 1	2021	202105	NA	1	5
Store 1	Client 1	2021	202106	NA	2	5
Store 1	Client 1	2021	202107	NA	3	5
Store 1	Client 1	2021	202108	TRUE	1	10
Store 1	Client 1	2021	202109	NA	4	5
Store 1	Client 1	2021	202110	NA	5	5

期望的结果是：

Store	Client	Year	Week	Exist	n_weekCount	maxWeek_Count
Store 1	Client 1	2020	202001	TRUE	1	8
Store 1	Client 1	2020	202002	TRUE	2	8
Store 1	Client 1	2020	202003	TRUE	3	8
Store 1	Client 1	2020	202004	TRUE	4	8
Store 1	Client 1	2020	202005	TRUE	5	8
Store 1	Client 1	2020	202006	TRUE	6	8
Store 1	Client 1	2020	202007	TRUE	7	8
Store 1	Client 1	2020	202008	TRUE	8	8
Store 1	Client 1	2020	202009	NA	1	3
Store 1	Client 1	2020	202010	NA	2	3
Store 1	Client 1	2020	202011	NA	3	3
Store 1	Client 1	2020	202012	TRUE	1	2
Store 1	Client 1	2020	202013	TRUE	2	2
Store 1	Client 1	2020	202014	NA	1	2
Store 1	Client 1	2020	202015	NA	2	2
Store 1	Client 1	2021	202101	TRUE	1	4
Store 1	Client 1	2021	202102	TRUE	2	4
Store 1	Client 1	2021	202103	TRUE	3	4
Store 1	Client 1	2021	202104	TRUE	4	4
Store 1	Client 1	2021	202105	NA	1	3
Store 1	Client 1	2021	202106	NA	2	3
Store 1	Client 1	2021	202107	NA	3	3
Store 1	Client 1	2021	202108	TRUE	1	1
Store 1	Client 1	2021	202109	NA	1	2
Store 1	Client 1	2021	202110	NA	2	2

【问题讨论】：

标签： r group-by count data.table

【解决方案1】：

我们可以使用 rleid 进行分组，并使用 seq_len(.N) 和组大小 (.N) 创建列 (:=)

library(data.table)
df[, c("n_WeekCount", "maxWeek_Count") := .(seq_len(.N), .N),
      .(grp = rleid(Exist), Store, Client, Year)]

-输出

> df
      Store   Client  Year   week  Exist n_WeekCount maxWeek_Count
     <char>   <char> <num>  <num> <char>       <int>         <int>
 1: store_1 client_1  2020 202001   TRUE           1             8
 2: store_1 client_1  2020 202002   TRUE           2             8
 3: store_1 client_1  2020 202003   TRUE           3             8
 4: store_1 client_1  2020 202004   TRUE           4             8
 5: store_1 client_1  2020 202005   TRUE           5             8
 6: store_1 client_1  2020 202006   TRUE           6             8
 7: store_1 client_1  2020 202007   TRUE           7             8
 8: store_1 client_1  2020 202008   TRUE           8             8
 9: store_1 client_1  2020 202009   <NA>           1             3
10: store_1 client_1  2020 202010   <NA>           2             3
11: store_1 client_1  2020 202011   <NA>           3             3
12: store_1 client_1  2020 202012   TRUE           1             2
13: store_1 client_1  2020 202013   TRUE           2             2
14: store_1 client_1  2020 202014   <NA>           1             2
15: store_1 client_1  2020 202015   <NA>           2             2
16: store_1 client_1  2021 202101   TRUE           1             4
17: store_1 client_1  2021 202102   TRUE           2             4
18: store_1 client_1  2021 202103   TRUE           3             4
19: store_1 client_1  2021 202104   TRUE           4             4
20: store_1 client_1  2021 202105   <NA>           1             3
21: store_1 client_1  2021 202106   <NA>           2             3
22: store_1 client_1  2021 202107   <NA>           3             3
23: store_1 client_1  2021 202108   TRUE           1             1
24: store_1 client_1  2021 202109   <NA>           1             2
25: store_1 client_1  2021 202110   <NA>           2             2

【讨论】：