獲取行索引以在循環中更新另一個數據幀

Question

我有以下數據：

編輯：

 df<- data.frame(
  id = c(432, 324, 322, 331, 242,443,223 ), 
  
  desc1= c("metal","steels&iron","irons\\copper", "sports material", "leather material", "durable goods", "electronic store")
  ,
  store_names = c("ik bros","steel idrs", "kb materials",  "ca pty (ltd)",  "bkk stores", "k/k \\shop", "h/j & jj")
  ,
  class = c("", "unknown","",  "sports", "unknown", "unknown", "") 
)

我想從 desc1 和 desc2 中搜索關鍵字，並為類列分配一個字符串值。 例如，關鍵字可以是

indus_1 <- c("iron", "steel")
goods_store_1 <- c("goods", "store", "stores")
electr_1 <- c("electronic", "chips", "semiconductor")
unlabelled_1 <- c("leather")

這里變量名 indus_1、sports_1 等將用於為類分配字符串值。 例如，如果找到“metal”關鍵字，我會在將“_1”剝離到 class 后分配 indus。 在我的方法中，我正在查找找到關鍵字的行的索引並將它們復制到相同數據幀的副本，但是對於較大的數據集，這需要很長時間，並且可能會錯過幾個類，因為我使用\\\\b來查找完全匹配。 這是預期的輸出：

  id            desc1  store_names       class
  432            metal      ik bros            
  324      steels&iron   steel idrs       indus
  322    irons\\copper kb materials       indus     
  331  sports material ca pty (ltd)      sports
  242 leather material   bkk stores  unlabelled
  443    durable goods   k/k \\shop goods_store
  223 electronic store     h/j & jj      electr

我正在尋找一種更有效的方法來做同樣的事情，一個完整的 dplyr 版本會更可取。 感謝您的建議。

Answer 1

不確定我是否正確解釋了您的問題； 這是你想做的嗎？

library(tidyverse)

df<- data.frame(
  id = c(432, 324, 322, 331, 242 ), 
  
  desc1 = c("iron and metal","sports material", "leather material", "durable goods", "electronic goods")
  ,
  desc2 = c("ik bros", "ca pty (ltd)",  "bkk stores", "k/k \\shop", "h/j & jj")
  ,
  class = c("", "sports", "unknown", "unknown", "") 
)

df2 <- df %>%
  mutate(class = case_when(str_detect(desc1, "metal") | str_detect(desc2, "metal")  ~ "indus",
                           str_detect(desc1, "sports") | str_detect(desc2, "sports") ~ "sports",
                           str_detect(desc1, "electronic") | str_detect(desc2, "electronic") ~ "electr",
                           str_detect(desc1, "goods") | str_detect(desc2, "goods") ~ "goods_store",
                           str_detect(desc1, "leather") | str_detect(desc2, "leather") ~ "unlabelled"))

df2
#>    id            desc1        desc2       class
#> 1 432   iron and metal      ik bros       indus
#> 2 324  sports material ca pty (ltd)      sports
#> 3 322 leather material   bkk stores  unlabelled
#> 4 331    durable goods   k/k \\shop goods_store
#> 5 242 electronic goods     h/j & jj      electr

^{由reprex 包(v2.0.1) 於 2021 年 10 月 25 日創建}

Answer 2

在這種情況下，你可以這樣做：

vars_1 <- mget(ls(pattern = '_1'))
vars_1 <- vars_1[!grepl('vars', names(vars_1))]
pat <- sub("_1", "", names(vars_1))
names(pat) <- sprintf(".*(%s).*", unlist(vars_1))

df %>%
  mutate(class = str_replace_all(invoke(str_c, across(starts_with('desc'))), pat))

  id            desc1        desc2       class
1 432   iron and metal      ik bros       indus
2 324  sports material ca pty (ltd)      sports
3 322 leather material   bkk stores  unlabelled
4 331    durable goods   k/k \\shop goods_store
5 242 electronic goods     h/j & jj      electr

Answer 3

從邏輯上講，我的答案類似於 @Onyambu 的答案，但幾乎沒有調整。

library(tidyverse)

mget(ls(pattern = '_1')) %>%
  stack() %>%
  group_by(ind = sub('_1', '', ind)) %>%
  summarise(values = sprintf('.*\\b(%s)\\b.*', paste0(values, collapse = '|'))) %>%
  select(2, 1) %>%
  deframe() -> pat

pat
#.*\\b(electronic|chips|semiconductor)\\b.*   .*\\b(goods|store|stores)\\b.*
#                                  "electr"           "goods_store" 
#                   .*\\b(iron|steel)\\b.*         .*\\b(leather)\\b.*
#                                   "indus"             "unlabelled"
df %>%
  mutate(class2 = str_replace_all(desc1, pat), 
         class2 = ifelse(desc1 == class2, '', class2))

#   id            desc1  store_names   class      class2
#1 432            metal      ik bros                    
#2 324      steels&iron   steel idrs unknown       indus
#3 322    irons\\copper kb materials                    
#4 331  sports material ca pty (ltd)  sports            
#5 242 leather material   bkk stores unknown  unlabelled
#6 443    durable goods   k/k \\shop unknown goods_store
#7 223 electronic store     h/j & jj              electr

對於id = 322它與indus不匹配，因為我們正在尋找完全匹配。 indus_1有iron而desc1列有irons 。