使用 r 基于组中的条件聚合从级别自动生成列答案

【问题标题】：Automatic column generation from levels based on conditional agggregation in groups using r使用 r 基于组中的条件聚合从级别自动生成列
【发布时间】：2021-10-12 23:11:02
【问题描述】：

对于下面的数据框，我正在尝试从 FactorCol1 有条件地创建八个额外的列 Last1Col7activ 到 Last10Col7inactive：

library(tidyverse)
Data_Frame <- data.frame(Col1 = c("A1", "A1", "A1", "A2", "A2", "A2", "A3", "A3", "A3"),
                         
                         Col2 = c("2011-03-11", "2014-08-21", "2016-01-17", "2017-06-30", "2018-07-11", "2018-11-28", "2019-09-04", "2020-02-29", "2020-07-12"),
                         
                         Col3 = c("2018-10-22", "2019-05-24", "2020-12-25", "2018-10-12", "2019-09-24", "2020-12-19", "2018-10-22", "2019-06-14", "2020-12-20"),
                         
                         Col4 = c(4, 2, 2, 1, 4, 4, 4, 4, 4),
                         
                         Col5 = c(7, 6, 3, 1, 3, 2, 5, 1, 2),
                         
                         FactorCol1 = c("active", "inactive", "inactive", "active", "active", "inactive", "inactive", "active", "inactive"),
                         
                         FactorCol2 = c("Level2", "Level2", "Level3", "Level1", "Level3", "Level1", "Level2", "Level1", "Level3"))

Data_Frame$Col1 <- as.factor(Data_Frame$Col1)
Data_Frame$Col2 <- as.Date(Data_Frame$Col2)
Data_Frame$Col3 <- as.Date(Data_Frame$Col3)
Data_Frame$FactorCol1 <- as.factor(Data_Frame$FactorCol1)
Data_Frame$FactorCol2 <- as.factor(Data_Frame$FactorCol2)

Data_Frame <- Data_Frame %>% group_by(Col1) %>% mutate(Col6 = lubridate::time_length(lubridate::interval(Col2, max(Col3)), "years"))
Data_Frame <- Data_Frame %>% group_by(Col1) %>% dplyr::mutate(Col7 = ifelse(Col6 <= 1, 1, ifelse(Col6 >1 & Col6 <=2, 2, ifelse(Col6 >2 & Col6 <=5, 5, ifelse(Col6 >5 & Col6 <=10, 10, 11)))))

Data_Frame <- Data_Frame %>% group_by(Col1) %>% dplyr::mutate(Col8 = ifelse(FactorCol1 == 'active', 1, 0))
Data_Frame <- Data_Frame %>% group_by(Col1) %>% dplyr::mutate(Col9 = ifelse(FactorCol1 == 'inactive', 1, 0))

Data_Frame <- as.data.frame(Data_Frame)

Data_Frame <- map_dfc(c(1, 2, 5, 10), ~ Data_Frame %>%
                          group_by(Col1) %>% 
                          transmute(!! sprintf("Last%dCol7active", .x) := sum(Col8[Col7 <= .x]),
                                    !! sprintf("Last%dCol7inactive", .x) := sum(Col9[Col7 <= .x])) %>% 
                          ungroup %>%
                          select(-Col1)) %>% 
bind_cols(Data_Frame, .)


  Col1       Col2       Col3 Col4 Col5 FactorCol1 FactorCol2      Col6 Col7 Col8 Col9 Last1Col7active Last1Col7inactive Last2Col7active
1   A1 2011-03-11 2018-10-22    4    7     active     Level2 9.7917808   10    1    0               0                 0               0
2   A1 2014-08-21 2019-05-24    2    6   inactive     Level2 6.3452055   10    0    1               0                 0               0
3   A1 2016-01-17 2020-12-25    2    3   inactive     Level3 4.9371585    5    0    1               0                 0               0
4   A2 2017-06-30 2018-10-12    1    1     active     Level1 3.4712329    5    1    0               0                 0               0
5   A2 2018-07-11 2019-09-24    4    3     active     Level3 2.4410959    5    1    0               0                 0               0
6   A2 2018-11-28 2020-12-19    4    2   inactive     Level1 2.0575342    5    0    1               0                 0               0
7   A3 2019-09-04 2018-10-22    4    5   inactive     Level2 1.2931507    2    0    1               1                 1               1
8   A3 2020-02-29 2019-06-14    4    1     active     Level1 0.8060109    1    1    0               1                 1               1
9   A3 2020-07-12 2020-12-20    4    2   inactive     Level3 0.4410959    1    0    1               1                 1               1
  Last2Col7inactive Last5Col7active Last5Col7inactive Last10Col7active Last10Col7inactive
1                 0               0                 1                1                  2
2                 0               0                 1                1                  2
3                 0               0                 1                1                  2
4                 0               2                 1                2                  1
5                 0               2                 1                2                  1
6                 0               2                 1                2                  1
7                 2               1                 2                1                  2
8                 2               1                 2                1                  2
9                 2               1                 2                1                  2

在哪里： Col6：每组内max(Col3)和Col2之间的时间差

Col7：Col6 中的值的 (

Col8：FactorCol1 中的活动元素设置为 1

Col9：FactorCol1 中的非活动元素设置为 1

Last1Col7active：在每个组内（Col1 中的 A1 到 A3），FactorCol1 中的活动元素数在 Col7 中

Last1Col7inactive：在每个组中，FactorCol1 中的非活动元素数量

Last5Col7active：在每个组内（Col1 中的 A1 到 A3），FactorCol1 中的活动元素数量

Last5Col7inactive：在每个组中，FactorCol1 中的非活动元素数在 Col7 中

Last10Col7active：在每个组内（Col1 中的 A1 到 A3），FactorCol1 中的活动元素数

Last10Col7inactive：在每个组内，FactorCol1 中的非活动元素数在 Col7 中

在尝试根据 FactorCol1 的级别自动生成列时，使用了以下代码，但是，结果显示 ....Col7inactive 的值始终被强制转换为 ...Col7active 的值。出了什么问题？

map_dfc(c(1, 2, 5, 10), function(.x) map_dfc(levels(Data_Frame$FactorCol1), function(.y) Data_Frame %>%
                                               group_by(Col1) %>%
                                               transmute(!! sprintf("Last%dCol7%s", .x, .y) := sum(Col8[Col7 <= .x])
                                                         ,!! sprintf("Last%dCol7%s", .x, .y) := sum(Col9[Col7 <= .x])
                                                         )%>% 
                                               ungroup %>%
                                               select(-Col1))) %>% 
bind_cols(Data_Frame, .)


  Col1       Col2       Col3 Col4 Col5 FactorCol1 FactorCol2      Col6 Col7 Col8 Col9 Last1Col7active Last1Col7inactive Last2Col7active
1   A1 2011-03-11 2018-10-22    4    7     active     Level2 9.7917808   10    1    0               0                 0               0
2   A1 2014-08-21 2019-05-24    2    6   inactive     Level2 6.3452055   10    0    1               0                 0               0
3   A1 2016-01-17 2020-12-25    2    3   inactive     Level3 4.9371585    5    0    1               0                 0               0
4   A2 2017-06-30 2018-10-12    1    1     active     Level1 3.4712329    5    1    0               0                 0               0
5   A2 2018-07-11 2019-09-24    4    3     active     Level3 2.4410959    5    1    0               0                 0               0
6   A2 2018-11-28 2020-12-19    4    2   inactive     Level1 2.0575342    5    0    1               0                 0               0
7   A3 2019-09-04 2018-10-22    4    5   inactive     Level2 1.2931507    2    0    1               1                 1               2
8   A3 2020-02-29 2019-06-14    4    1     active     Level1 0.8060109    1    1    0               1                 1               2
9   A3 2020-07-12 2020-12-20    4    2   inactive     Level3 0.4410959    1    0    1               1                 1               2
  Last2Col7inactive Last5Col7active Last5Col7inactive Last10Col7active Last10Col7inactive
1                 0               1                 1                2                  2
2                 0               1                 1                2                  2
3                 0               1                 1                2                  2
4                 0               1                 1                1                  1
5                 0               1                 1                1                  1
6                 0               1                 1                1                  1
7                 2               2                 2                2                  2
8                 2               2                 2                2                  2
9                 2               2                 2                2                  2

【问题讨论】：

标签： r dataframe group-by conditional-statements tidyverse

【解决方案1】：

我想我通过以下代码得到了预期的输出：

library(tidyverse)

 # 1st step, make the changes to colum types and create Col7 tp 9
df <- df %>% 
  # Here I just make the changes to colum format as you did in your code ex
  mutate(
    across(.cols = any_of(c("Col1", "FactorCol1", "FactorCol2")), 
           .fns = as.factor),
    across(.cols = any_of(c("Col2", "Col3")),
           .fns = as.Date)
    
  ) %>% 
  group_by(Col1) %>% 
  mutate(
    Col6 = lubridate::time_length(lubridate::interval(Col2, max(Col3)), "years"),
    Col7 = ifelse(Col6 <= 1, 1, ifelse(Col6 >1 & Col6 <=2, 2, ifelse(Col6 >2 & Col6 <=5, 5, ifelse(Col6 >5 & Col6 <=10, 10, 11)))),
    Col8 = ifelse(FactorCol1 == 'active', 1, 0),
    Col9 = ifelse(FactorCol1 == 'inactive', 1, 0),
  )
# make the counts and merge them with initial df
df %>% 
  group_by(Col1, FactorCol1) %>% 
  # the trick is to use summarise to get the counts and then use pivot_wider and merge to "set that entire column to those number"
  summarise(
    Last1Col7 = sum(Col7 <= 1),
    Last5Col7 = sum(Col7 <= 5),
    Last10Col7 = sum(Col7 <= 10)
  ) %>% 
  pivot_wider(names_from = FactorCol1, values_from = 3:last_col(), names_sep = "") %>% 
  right_join(df, ., by = "Col1")

数据：

df <- data.frame(Col1 = c("A1", "A1", "A1", "A2", "A2", "A2", "A3", "A3", "A3"),
                         
                         Col2 = c("2011-03-11", "2014-08-21", "2016-01-17", "2017-06-30", "2018-07-11", "2018-11-28", "2019-09-04", "2020-02-29", "2020-07-12"),
                         
                         Col3 = c("2018-10-22", "2019-05-24", "2020-12-25", "2018-10-12", "2019-09-24", "2020-12-19", "2018-10-22", "2019-06-14", "2020-12-20"),
                         
                         Col4 = c(4, 2, 2, 1, 4, 4, 4, 4, 4),
                         
                         Col5 = c(7, 6, 3, 1, 3, 2, 5, 1, 2),
                         
                         FactorCol1 = c("active", "inactive", "inactive", "active", "active", "inactive", "inactive", "active", "inactive"),
                         
                         FactorCol2 = c("Level2", "Level2", "Level3", "Level1", "Level3", "Level1", "Level2", "Level1", "Level3"))

进一步解释回答“为什么要设置pivot_wider(values_from = 3:last_col())？”

df %>% group_by(Col1, FactorCol1) %>% summarise(
    Last1Col7 = sum(Col7 <= 1),
    Last5Col7 = sum(Col7 <= 5),
    Last10Col7 = sum(Col7 <= 10))

给你以下小标题：

# A tibble: 6 x 5
# Groups:   Col1 [3]
  Col1  FactorCol1 Last1Col7 Last5Col7 Last10Col7
  <fct> <fct>          <int>     <int>      <int>
1 A1    active             0         0          1
2 A1    inactive           0         1          2
3 A2    active             0         2          2
4 A2    inactive           0         1          1
5 A3    active             1         1          1
6 A3    inactive           1         2          2

您希望FactorCol1 的信息采用宽格式，新列包含从第三列 (Last1Col7) 到最后一列的值。从?pivot_wider，我们知道 "names_from, values_from : 一对参数，描述从哪一列（或多列）获取输出列的名称（names_from），以及从哪一列（或多列）获取单元格values from (values_from)。如果 values_from 包含多个值，则该值将添加到输出列的前面。"

然后，使用%>% pivot_wider(names_from = FactorCol1, values_from = 3:last_col(), names_sep = "")，你会得到：

# A tibble: 3 x 7
# Groups:   Col1 [3]
  Col1  Last1Col7active Last1Col7inactive Last5Col7active Last5Col7inactive Last10Col7active Last10Col7inactive
  <fct>           <int>             <int>           <int>             <int>            <int>              <int>
1 A1                  0                 0               0                 1                1                  2
2 A2                  0                 0               2                 1                2                  1
3 A3                  1                 1               1                 2                1                  2

【讨论】：

谢谢。能否请您解释 values_from = 3 的作用：（以及为什么是 3）？
我运行df %>% group_by(Col1, FactorCol1) %>% # the trick is to use summarise to get the counts and then use pivot_wider and merge to "set that entire column to those number" summarise( Last1Col7 = sum(Col7 <= 1), Last5Col7 = sum(Col7 <= 5), Last10Col7 = sum(Col7 <= 10) )，您将在控制台中看到数据框的状态。要将FactorCol1 的级别作为列，您需要枢轴是。并且您希望保留最后 3 列的值。
我希望它更清楚^^顺便说一句，请参阅更新的答案。我虽然可以在一个 dplyr 链中完成所有操作，但我觉得需要保存一个中间 df 以保留您在最终合并中创建的列。
谢谢，我仍然没有理解与 3 相关的要点（它们是最后 3 列）。
@Ray 答案已编辑并进一步解释。

【解决方案2】：

以下内容可以完成这项工作，但它相当于创建额外的列并删除它们，但使用其他列名

ABC <- map_dfc(c(1, 2, 5, 10), function(.x) map_dfc(levels(Data_Frame$FactorCol1), function(.y) Data_Frame %>%
                                                             group_by(Col1) %>%
                                                             transmute(!! sprintf("Last%dCol7Col8%s", .x, .y) := sum(Col8[Col7 <= .x]),
                                                                       !! sprintf("Last%dCol7Col9%s", .x, .y) := sum(Col9[Col7 <= .x])
                                                             )%>% 
                                                             ungroup %>%
                                                             select(-Col1))) %>%
  bind_cols(Data_Frame, .)

# removing all columns that contain these names
Data_Frame <- Data_Frame %>% select(-contains(c("Col7Col9active", "Col7Col8inactive")))

【讨论】：