解析 R 中的字符串并将值与另一列进行比较

Question

这可能看起来微不足道，但我真的陷入了将值与这个复杂字符串进行比较的问题

我的数据框如下所示：

Id	History	Report Month
1001	Jun:2020,030/XXX-May:2020,035/XXX-Apr:2020,040/XXX-Mar:2020,060/XXX	July 2021
1003	Jun:2017,823/XXX-May:2017,000/XXX-Apr:2017,000/XXX-Mar:2017,000/XXX	July 2021
1005	Apr:2019,000/XXX-Mar:2019,800/XXX-Feb:2019,000/XXX-Jan:2019,000/XXX	July 2021
1006	Jun:2020,000/XXX-May:2020,030/XXX-Apr:2020,060/XXX-Mar:2020,090/XXX	July 2021

用于比较的列历史的键、值对如下：

Id : 1001 - Jun 2020,030   May 2020, 035   Apr 2020, 040...... 
Id : 1003 - Jun 2017,823   May 2017, 000   Apr 2017, 000...... 

问题陈述是：我想将这些键、值对与报告月份（即总是当前月份）进行比较，并根据它创建一个条件列。逻辑是：2021 年 7 月之前的 24 个月（可能是 12 或 36 个月），即 2021 年 7 月至 2019 年 6 月，在此时间段内的月份中有多少键、值对的值 >= 30 或 >= 60 等。所以如果一个字符串从

输出

Id	Report Month	+30_last_24	+30_last_36
1001	July 2021	4	4
1003	July 2021	0	0
1005	July 2021	0	1
1006	July 2021	3	3

我最近才开始使用 R，甚至没有解决方案，所以任何帮助都将不胜感激。

修改后的原始数据集

df <- read.table(header = T, text = "Id History ReportMonth
1001    Jun:2020,030/XXX|May:2020,035/XXX|Apr:2020,040/XXX|Mar:2020,060/XXX 'July 2021'
1003    Jun:2017,DDD/XXX|May:2017,030/XXX|Apr:2017,DDD/STD|Mar:2017,000/XXX 'July 2021'
1005    Apr:2019,000/XXX|Mar:2019,800/DDD|Feb:2019,000/XXX|Jan:2019,000/XXX 'July 2021'
1006    Jun:2020,000/XXX|May:2020,030/XXX|Apr:2020,060/XXX|Mar:2020,090/XXX 'July 2021'")

Answer 1

根据修改修订战略-

• 使用| 分隔行，但仅在使用\\ 转义之后
• 使用, 分隔成列
• 使用gsub从值中提取数字
• 休息很明显。

如有任何疑问，请随时询问。

df <- read.table(header = T, text = "Id History ReportMonth
1001    Jun:2020,030/XXX|May:2020,035/XXX|Apr:2020,040/XXX|Mar:2020,060/XXX 'July 2021'
1003    Jun:2017,DDD/XXX|May:2017,030/XXX|Apr:2017,DDD/STD|Mar:2017,000/XXX 'July 2021'
1005    Apr:2019,000/XXX|Mar:2019,800/DDD|Feb:2019,000/XXX|Jan:2019,000/XXX 'July 2021'
1006    Jun:2020,000/XXX|May:2020,030/XXX|Apr:2020,060/XXX|Mar:2020,090/XXX 'July 2021'")

library(tidyverse)
library(lubridate, warn.conflicts = F)

df %>%
  separate_rows(History, sep = '\\|') %>%
  separate(History, into = c('Hist_mon', 'Hist_val'), sep = ',') %>%
  mutate(Hist_mon = dmy(paste0('1:', Hist_mon)),
         Hist_val = as.numeric(gsub('(\\D*)', '', Hist_val)),
         ReportMonth = dmy(paste0('1 ', ReportMonth))) %>%
  group_by(Id, ReportMonth) %>%
  summarise(last_24_30 = sum(Hist_val >= 30 & Hist_mon >= ReportMonth %m-% months(24)),
            last_36_30 = sum(Hist_val >= 30 & Hist_mon >= ReportMonth %m-% months(36)), .groups = 'drop')
#> # A tibble: 4 x 4
#>      Id ReportMonth last_24_30 last_36_30
#>   <int> <date>           <int>      <int>
#> 1  1001 2021-07-01           4          4
#> 2  1003 2021-07-01           0          0
#> 3  1005 2021-07-01           0          1
#> 4  1006 2021-07-01           3          3

^{由reprex package 创建于 2021-07-16 (v2.0.0)}

Answer 2

library(tidyverse)
library(lubridate)
df %>%
  separate_rows(History, sep = '[|]')%>%
  filter(str_detect(History, "\\w"), str_detect(History, "\\d+/"))%>%
  separate(History, c("Date", "Value", "d"), sep = '[,/]', convert = TRUE) %>%
  mutate(across(c(Date,ReportMonth), ~myd(paste(.x, "01")))) %>%
  group_by(Id) %>%
  summarise(r = list(map(c(m24 = 24, m36 = 36), ~sum(
    Date + months(.x) > ReportMonth & Value >= 30)))) %>%
  unnest_wider(r) %>%
  right_join(df, 'Id')

  
# A tibble: 4 x 5
     Id   m24   m36 History_Report                                                      Month    
  <int> <int> <int> <chr>                                                               <chr>    
1  1001     4     4 Jun:2020,030/XXX-May:2020,035/XXX-Apr:2020,040/XXX-Mar:2020,060/XXX July 2021
2  1003     0     0 Jun:2017,823/XXX-May:2017,000/XXX-Apr:2017,000/XXX-Mar:2017,000/XXX July 2021
3  1005     0     1 Apr:2019,000/XXX-Mar:2019,800/XXX-Feb:2019,000/XXX-Jan:2019,000/XXX July 2021
4  1006     3     3 Jun:2020,000/XXX-May:2020,030/XXX-Apr:2020,060/XXX-Mar:2020,090/XXX July 2021

---