R - 如何根据 open/close 帐户日期计算用户数，但用户有多个帐户
R - how to count users based on open/close dates of accounts, but with users having multiple accounts
我有一个帐户列表（超过 30 万行），可以追溯到六年前，其中包含用户编号、打开和关闭日期以及其他信息，例如位置。我们提供多种账户，一个用户可以拥有一个或多个账户，可以任意组合，可以连续也可以重叠。
有人要求我了解我们在任何给定月份有多少用户。他们希望它按位置和总数分开。
所以我有一个 table 这样的：
User Open Close Area
1 A 2018-02-13 2018-07-31 West
2 B 2018-02-26 2018-06-04 North
3 B 2018-02-27 2018-03-15 North
4 C 2018-02-27 2018-05-26 South
5 C 2018-03-15 2018-06-03 South
6 D 2018-03-20 2018-07-02 East
7 E 2018-04-01 2018-06-19 West
8 E 2018-04-14 2018-05-04 West
9 F 2018-03-20 2018-04-19 North
10 G 2018-04-26 2018-07-04 South
11 H 2017-29-12 2018-03-21 East
12 I 2016-11-29 2020-04-10 West
13 J 2018-01-31 2018-12-20 West
14 K 2017-10-31 2018-10-30 North
15 K 2018-10-31 2019-10-30 North
我想要一个看起来像这样的人：
Month Total North East South West
1 Feb 18 3 1 0 1 1
2 Mar 18 5 2 1 1 1
3 Apr 18 7 2 1 2 2
4 May 18 6 1 1 2 2
5 Jun 18 6 1 1 2 2
6 Jul 18 3 0 1 1 1
我可以使用
过滤数据以获得我需要的个别月份的数据
df%>%
filter(Open <= as.Date("2018-04-30") & Close >= as.Date("2018-04-01")) %>%
distinct(PERSON_ID, .keep_all = TRUE) %>%
count(Area)
但我想不通的是如何在数据集中自动重复每个月。有什么地方可以让 r 在我的数据集中每个月重复上述操作，然后将结果传递给第二个 table?
非常感谢您提供的所有帮助，非常感谢您的宝贵时间。
编辑：在 Matin Gal 的解决方案多年返回 NA 的源数据中添加示例
这是我的做法：
library(tidyverse)
set.seed(14159)
## generating some data that looks roughly
## like your data
data <- tibble(
user = sample(LETTERS[1:5], size = 20, replace = TRUE),
open = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 20),
close = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 20),
area = sample(c("N", "E", "S", "W"), 20, replace = T)
) %>%
filter(
close > open
)
data
#> # A tibble: 9 × 4
#> user open close area
#> <chr> <date> <date> <chr>
#> 1 A 1999-04-03 1999-07-28 N
#> 2 B 1999-01-27 1999-05-12 W
#> 3 B 1999-06-05 1999-12-29 W
#> 4 C 1999-09-26 1999-12-30 W
#> 5 C 1999-04-21 1999-12-04 E
#> 6 C 1999-08-11 1999-12-12 N
#> 7 A 1999-02-13 1999-09-16 W
#> 8 E 1999-02-17 1999-05-21 E
#> 9 B 1999-07-26 1999-08-16 S
## figuring out what months are in between open and close
get_months_in_range <- function(open, close) {
seq.Date(
open,
close,
by = "month"
) %>%
list()
}
data %>%
rowwise() %>%
mutate(
Month = get_months_in_range(open, close)
) %>%
ungroup() %>%
unnest_longer(
col = Month
) %>%
count(Month, area) %>%
pivot_wider(
names_from = area,
values_from = n,
values_fill = 0
) %>%
rowwise() %>%
mutate(
Total = sum(
c_across(
-Month
)
)
) %>%
ungroup()
#> # A tibble: 45 × 6
#> Month W E N S Total
#> <date> <int> <int> <int> <int> <int>
#> 1 1999-01-27 1 0 0 0 1
#> 2 1999-02-13 1 0 0 0 1
#> 3 1999-02-17 0 1 0 0 1
#> 4 1999-02-27 1 0 0 0 1
#> 5 1999-03-13 1 0 0 0 1
#> 6 1999-03-17 0 1 0 0 1
#> 7 1999-03-27 1 0 0 0 1
#> 8 1999-04-03 0 0 1 0 1
#> 9 1999-04-13 1 0 0 0 1
#> 10 1999-04-17 0 1 0 0 1
#> # … with 35 more rows
由 reprex package (v2.0.1)
于 2021-08-18 创建
这不是世界上最性感的解决方案，但我认为它会带您到达您想去的地方。基本上，我只是制作了一个辅助函数，它为我提供了 open 和 close 之间的所有日期，然后您可以将这些日期分组以计算出您在任何给定月份有多少用户。如果您想了解有关 dplyr 长链正在做什么的更多解释，请告诉我。
欢迎来到 SO。我无法测试此代码，因为您没有以正确的格式提供数据片段（有关这一点的建议，请参见下文），但我认为您想要做的基本想法是提取一个月-来自 Open 的年份值，然后使用 group_by。例如：
library(lubridate)
library(dplyr)
df %>% mutate(
Date = dmy(Open),
Month_Yr = format_ISO8601(Date, precision = "ym")) %>%
group_by(Month_Yr) %>%
distinct(PERSON.ID, .keep_all = TRUE) %>%
count(Area)
通常在 SO 上共享数据时最好使用 dput。如果您不确定，请参阅 ?dput 了解如何使用它。
这是适用于超过一年的日期的通用解决方案。
library(dplyr)
library(tidyr)
library(lubridate)
data %>%
group_by(rn = row_number()) %>%
mutate(seq = list(seq(month(Open), month(Close) + 12 * (year(Close) - year(Open))))) %>%
unnest(seq) %>%
mutate(
seq_2 = (seq - 1) %% 12 + 1,
month = month(seq_2, label = TRUE),
year = year(Open + months(seq - first(seq)))
) %>%
ungroup() %>%
distinct(User, month, year, Area) %>%
count(month, year, Area) %>%
pivot_wider(
names_from = "Area",
values_from = "n",
values_fill = 0
) %>%
mutate(Total = rowSums(across(c(North, South, West, East))))
returns
month year North South West East Total
<ord> <dbl> <int> <int> <int> <int> <dbl>
1 Feb 2018 1 1 1 0 3
2 Mar 2018 2 1 1 1 5
3 Apr 2018 2 2 2 1 7
4 May 2018 1 2 2 1 6
5 Jun 2018 1 2 2 1 6
6 Jul 2018 0 1 1 1 3
数据
df <- structure(list(id = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), User = c("A",
"B", "B", "C", "C", "D", "E", "E", "F", "G"), Open = structure(c(17575,
17588, 17589, 17589, 17605, 17610, 17622, 17635, 17610, 17647
), class = "Date"), Close = structure(c(17743, 17686, 17605,
17677, 17685, 17714, 17701, 17655, 17640, 17716), class = "Date"),
Area = c("West", "North", "North", "South", "South", "East",
"West", "West", "North", "South")), problems = structure(list(
row = 10L, col = "Area", expected = "", actual = "embedded null",
file = "literal data"), row.names = c(NA, -1L), class = c("tbl_df",
"tbl", "data.frame")), class = c("spec_tbl_df", "tbl_df", "tbl",
"data.frame"), row.names = c(NA, -10L), spec = structure(list(
cols = list(id = structure(list(), class = c("collector_double",
"collector")), User = structure(list(), class = c("collector_character",
"collector")), Open = structure(list(format = ""), class = c("collector_date",
"collector")), Close = structure(list(format = ""), class = c("collector_date",
"collector")), Area = structure(list(), class = c("collector_character",
"collector"))), default = structure(list(), class = c("collector_guess",
"collector")), skip = 1L), class = "col_spec"))
我有一个帐户列表（超过 30 万行），可以追溯到六年前，其中包含用户编号、打开和关闭日期以及其他信息，例如位置。我们提供多种账户，一个用户可以拥有一个或多个账户，可以任意组合，可以连续也可以重叠。
有人要求我了解我们在任何给定月份有多少用户。他们希望它按位置和总数分开。
所以我有一个 table 这样的：
User Open Close Area
1 A 2018-02-13 2018-07-31 West
2 B 2018-02-26 2018-06-04 North
3 B 2018-02-27 2018-03-15 North
4 C 2018-02-27 2018-05-26 South
5 C 2018-03-15 2018-06-03 South
6 D 2018-03-20 2018-07-02 East
7 E 2018-04-01 2018-06-19 West
8 E 2018-04-14 2018-05-04 West
9 F 2018-03-20 2018-04-19 North
10 G 2018-04-26 2018-07-04 South
11 H 2017-29-12 2018-03-21 East
12 I 2016-11-29 2020-04-10 West
13 J 2018-01-31 2018-12-20 West
14 K 2017-10-31 2018-10-30 North
15 K 2018-10-31 2019-10-30 North
我想要一个看起来像这样的人：
Month Total North East South West
1 Feb 18 3 1 0 1 1
2 Mar 18 5 2 1 1 1
3 Apr 18 7 2 1 2 2
4 May 18 6 1 1 2 2
5 Jun 18 6 1 1 2 2
6 Jul 18 3 0 1 1 1
我可以使用
过滤数据以获得我需要的个别月份的数据 df%>%
filter(Open <= as.Date("2018-04-30") & Close >= as.Date("2018-04-01")) %>%
distinct(PERSON_ID, .keep_all = TRUE) %>%
count(Area)
但我想不通的是如何在数据集中自动重复每个月。有什么地方可以让 r 在我的数据集中每个月重复上述操作，然后将结果传递给第二个 table?
非常感谢您提供的所有帮助，非常感谢您的宝贵时间。
编辑：在 Matin Gal 的解决方案多年返回 NA 的源数据中添加示例
这是我的做法：
library(tidyverse)
set.seed(14159)
## generating some data that looks roughly
## like your data
data <- tibble(
user = sample(LETTERS[1:5], size = 20, replace = TRUE),
open = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 20),
close = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 20),
area = sample(c("N", "E", "S", "W"), 20, replace = T)
) %>%
filter(
close > open
)
data
#> # A tibble: 9 × 4
#> user open close area
#> <chr> <date> <date> <chr>
#> 1 A 1999-04-03 1999-07-28 N
#> 2 B 1999-01-27 1999-05-12 W
#> 3 B 1999-06-05 1999-12-29 W
#> 4 C 1999-09-26 1999-12-30 W
#> 5 C 1999-04-21 1999-12-04 E
#> 6 C 1999-08-11 1999-12-12 N
#> 7 A 1999-02-13 1999-09-16 W
#> 8 E 1999-02-17 1999-05-21 E
#> 9 B 1999-07-26 1999-08-16 S
## figuring out what months are in between open and close
get_months_in_range <- function(open, close) {
seq.Date(
open,
close,
by = "month"
) %>%
list()
}
data %>%
rowwise() %>%
mutate(
Month = get_months_in_range(open, close)
) %>%
ungroup() %>%
unnest_longer(
col = Month
) %>%
count(Month, area) %>%
pivot_wider(
names_from = area,
values_from = n,
values_fill = 0
) %>%
rowwise() %>%
mutate(
Total = sum(
c_across(
-Month
)
)
) %>%
ungroup()
#> # A tibble: 45 × 6
#> Month W E N S Total
#> <date> <int> <int> <int> <int> <int>
#> 1 1999-01-27 1 0 0 0 1
#> 2 1999-02-13 1 0 0 0 1
#> 3 1999-02-17 0 1 0 0 1
#> 4 1999-02-27 1 0 0 0 1
#> 5 1999-03-13 1 0 0 0 1
#> 6 1999-03-17 0 1 0 0 1
#> 7 1999-03-27 1 0 0 0 1
#> 8 1999-04-03 0 0 1 0 1
#> 9 1999-04-13 1 0 0 0 1
#> 10 1999-04-17 0 1 0 0 1
#> # … with 35 more rows
由 reprex package (v2.0.1)
于 2021-08-18 创建这不是世界上最性感的解决方案，但我认为它会带您到达您想去的地方。基本上，我只是制作了一个辅助函数，它为我提供了 open 和 close 之间的所有日期，然后您可以将这些日期分组以计算出您在任何给定月份有多少用户。如果您想了解有关 dplyr 长链正在做什么的更多解释，请告诉我。
欢迎来到 SO。我无法测试此代码，因为您没有以正确的格式提供数据片段（有关这一点的建议，请参见下文），但我认为您想要做的基本想法是提取一个月-来自 Open 的年份值，然后使用 group_by。例如：
library(lubridate)
library(dplyr)
df %>% mutate(
Date = dmy(Open),
Month_Yr = format_ISO8601(Date, precision = "ym")) %>%
group_by(Month_Yr) %>%
distinct(PERSON.ID, .keep_all = TRUE) %>%
count(Area)
通常在 SO 上共享数据时最好使用 dput。如果您不确定，请参阅 ?dput 了解如何使用它。
这是适用于超过一年的日期的通用解决方案。
library(dplyr)
library(tidyr)
library(lubridate)
data %>%
group_by(rn = row_number()) %>%
mutate(seq = list(seq(month(Open), month(Close) + 12 * (year(Close) - year(Open))))) %>%
unnest(seq) %>%
mutate(
seq_2 = (seq - 1) %% 12 + 1,
month = month(seq_2, label = TRUE),
year = year(Open + months(seq - first(seq)))
) %>%
ungroup() %>%
distinct(User, month, year, Area) %>%
count(month, year, Area) %>%
pivot_wider(
names_from = "Area",
values_from = "n",
values_fill = 0
) %>%
mutate(Total = rowSums(across(c(North, South, West, East))))
returns
month year North South West East Total
<ord> <dbl> <int> <int> <int> <int> <dbl>
1 Feb 2018 1 1 1 0 3
2 Mar 2018 2 1 1 1 5
3 Apr 2018 2 2 2 1 7
4 May 2018 1 2 2 1 6
5 Jun 2018 1 2 2 1 6
6 Jul 2018 0 1 1 1 3
数据
df <- structure(list(id = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), User = c("A",
"B", "B", "C", "C", "D", "E", "E", "F", "G"), Open = structure(c(17575,
17588, 17589, 17589, 17605, 17610, 17622, 17635, 17610, 17647
), class = "Date"), Close = structure(c(17743, 17686, 17605,
17677, 17685, 17714, 17701, 17655, 17640, 17716), class = "Date"),
Area = c("West", "North", "North", "South", "South", "East",
"West", "West", "North", "South")), problems = structure(list(
row = 10L, col = "Area", expected = "", actual = "embedded null",
file = "literal data"), row.names = c(NA, -1L), class = c("tbl_df",
"tbl", "data.frame")), class = c("spec_tbl_df", "tbl_df", "tbl",
"data.frame"), row.names = c(NA, -10L), spec = structure(list(
cols = list(id = structure(list(), class = c("collector_double",
"collector")), User = structure(list(), class = c("collector_character",
"collector")), Open = structure(list(format = ""), class = c("collector_date",
"collector")), Close = structure(list(format = ""), class = c("collector_date",
"collector")), Area = structure(list(), class = c("collector_character",
"collector"))), default = structure(list(), class = c("collector_guess",
"collector")), skip = 1L), class = "col_spec"))