专门使用 grepl 有条件地提取多个子字符串，索引提取以返回累积值

Question

祝大家今天好，

我无法完成这项具有挑战性的任务，因此我想找到一种优雅的方法来：

1. 我需要对“区域”中的每个行元素使用一种适应性强的方法，例如循环
2. 从按“Zone”元素分组的“country_name”中逐行提取多个子字符串
3. 将逐行的多个子字符串存储为索引值以用于 df2
4. 将索引值与 df2 中的数据框匹配
5. 计算总人口并根据 df1 对其进行变异

本质上的挑战是，该方法不应该针对数据框中的任何特定元素进行修复。

第一个数据帧：

df1 <- data.frame(zone, country_name)
zone = c("M", "N", "O")
country_name = c("The USA, Canada & Mexico are part of North America", "Canada like Australia is a Commonwealth member", "The UK is still finalizing its exit plans from the EU")

第二个数据框：

df2 <- data.frame(zonal_region, country, population)
zonal_region = c("M", "M", "M", "N", "N", "N", "O", "O", "O")
country = c("USA", "Canada", "Mexico", "Canada", "Australia", "UK", "Australia", "UK", "Canada")
population = c(323.4 , 36.29, 127.5, 36.29, 24.13, 65.64, 24.13, 65.64, 36.29)

这是我最终输出的样子：

df3 <- data.frame(zone, country_name, total_population)
zone = c("M", "N", "O")
country_name = c("The USA, Canada & Mexico are part of North America", "Canada like Australia is a Commonwealth member", "The UK is still finalizing its exit plans from the EU")
total_population = c(487.19, 60.42, 65.64)

我在提取多个子字符串并针对给定区域的 df2 索引它们的值时遇到了麻烦。

如果有人能解决这个问题，将不胜感激。

谢谢！

Answer 1

你可以试试fuzzyjoin

library(dplyr)
library(stringr)
library(fuzzyjoin)

df1 %>% 
  mutate_if(is.factor, as.character) %>%
  fuzzy_left_join((df2 %>% mutate_if(is.factor, as.character)),
                  by = c("zone" = "zonal_region", "country_name" = "country"), 
                  match_fun = str_detect) %>%
  group_by(zone, country_name) %>%
  summarise(total_population = sum(population)) %>%
  data.frame()

输出为：

  zone                                          country_name total_population
1    M    The USA, Canada & Mexico are part of North America           487.19
2    N        Canada like Australia is a Commonwealth member            60.42
3    O The UK is still finalizing its exit plans from the EU            65.64

样本数据：

df1 <- structure(list(zone = structure(1:3, .Label = c("M", "N", "O"
), class = "factor"), country_name = structure(c(3L, 1L, 2L), .Label = c("Canada like Australia is a Commonwealth member", 
"The UK is still finalizing its exit plans from the EU", "The USA, Canada & Mexico are part of North America"
), class = "factor")), class = "data.frame", row.names = c(NA, 
-3L))

df2 <- structure(list(zonal_region = structure(c(1L, 1L, 1L, 2L, 2L, 
2L, 3L, 3L, 3L), .Label = c("M", "N", "O"), class = "factor"), 
    country = structure(c(5L, 2L, 3L, 2L, 1L, 4L, 1L, 4L, 2L), .Label = c("Australia", 
    "Canada", "Mexico", "UK", "USA"), class = "factor"), population = c(323.4, 
    36.29, 127.5, 36.29, 24.13, 65.64, 24.13, 65.64, 36.29)), class = "data.frame", row.names = c(NA, 
-9L))

Answer 2

我们可以在从 'df1' 的 'country_name' 列中提取 'country' 并执行 group_by sum 后对两个数据集执行 left/right joins 来做到这一点

library(tidyverse)
un1 <- unique(df2$country)
df1 %>%
   mutate(cntry =  str_extract_all(country_name, paste(un1, collapse="|"))) %>% 
   right_join(df2, by = c('zone' = 'zonal_region')) %>% 
   group_by(zone) %>% 
   summarize(total_population= sum(population[country %in% cntry[[1]]])) %>% 
   left_join(df1) %>%
   select(zone, country_name, total_population)
# A tibble: 3 x 3
#  zone  country_name                                          total_population
  <fct> <fct>                                                            <dbl>
#1 M     The USA, Canada & Mexico are part of North America               487. 
#2 N     Canada like Australia is a Commonwealth member                    60.4
#3 O     The UK is still finalizing its exit plans from the EU             65.6