基于对象状态的条件累积和

Question

我正在寻找基于对象状态的条件累计和。当统计数据为“新”时，我想对所有前面的行求和，但当状态变为“旧”时重置累计总和。我还想对组 ID 进行迭代。

因此，在下面的示例中：

set.seed(5)
df <- data.table(ID=c(rep("A",10),rep("B",10)),time=c(seq(1,10),seq(1,10)),
                 Status=sample(c("New","Old"),20,replace = TRUE))
df
    ID time Status
 1:  A    1    Old
 2:  A    2    New
 3:  A    3    New
 4:  A    4    New
 5:  A    5    New
 6:  A    6    New
 7:  A    7    New
 8:  A    8    New
 9:  A    9    Old
10:  A   10    New
11:  B    1    New
12:  B    2    New
13:  B    3    New
14:  B    4    Old
15:  B    5    Old
16:  B    6    New
17:  B    7    New
18:  B    8    Old
19:  B    9    Old
20:  B   10    Old

期望的结果是：

    ID time Status Cond_Sum
 1:  A    1    Old        0
 2:  A    2    New        1
 3:  A    3    New        2
 4:  A    4    New        3
 5:  A    5    New        4
 6:  A    6    New        5
 7:  A    7    New        6
 8:  A    8    New        7
 9:  A    9    Old        0
10:  A   10    New        1
11:  B    1    New        1
12:  B    2    New        2
13:  B    3    New        3
14:  B    4    Old        0
15:  B    5    Old        0
16:  B    6    New        1
17:  B    7    New        2
18:  B    8    Old        0
19:  B    9    Old        0
20:  B   10    Old        0

首选data.table解决方案。

非常感谢。

Answer 1

我们可以在'ID'、'Status'上创建一个rleid的分组列，然后在i(Status == "New")中指定条件表达式，将行序列 (seq_len(.N)) 分配给按 'grp' 分组的 'Cond_Sum'（或使用 rowid(grp)）

library(data.table)
df[, grp := rleid(ID, Status)]
df[, Cond_Sum := 0][Status == 'New', 
          Cond_Sum := seq_len(.N), grp][, grp := NULL][]

-输出

#     ID time Status Cond_Sum
# 1:  A    1    Old        0
# 2:  A    2    New        1
# 3:  A    3    New        2
# 4:  A    4    New        3
# 5:  A    5    New        4
# 6:  A    6    New        5
# 7:  A    7    New        6
# 8:  A    8    New        7
# 9:  A    9    Old        0
#10:  A   10    New        1
#11:  B    1    New        1
#12:  B    2    New        2
#13:  B    3    New        3
#14:  B    4    Old        0
#15:  B    5    Old        0
#16:  B    6    New        1
#17:  B    7    New        2
#18:  B    8    Old        0
#19:  B    9    Old        0
#20:  B   10    Old        0

基于对象状态的条件累积和

Conditional Cumulative sum based on the status of an object

iteration

r

cumsum

data.table