使用 dplyr 按列类别将具有多行值的数据框重组为单行

Question

以下是 dataframe 子集的示例我在 R有多个公司的信息超过多个 rows按类别 - company_name , no_workers, product , address和 contact_person )

comp_df <- structure(list(desc = c("AAA", "Company", "Ltd", "fish", "344", 
"12", "West", "Road", "Bob C", "BBB", "Enteprises", "vegetables", 
"12", "North", "Perak", "Simon T", "EF", "Industries", "cement", 
"8800", "Green", "Lane", "Singapore", "Sylvia P"), category = c("company_name", 
"company_name", "company_name", "product", "no_workers", "address", 
"address", "address", "contact_person", "company_name", "company_name", 
"product", "no_workers", "address", "address", "contact_person", 
"company_name", "company_name", "product", "no_workers", "address", 
"address", "address", "contact_person")), row.names = c(NA, -24L
), class = c("tbl_df", "tbl", "data.frame"))

有没有简单的方法可以添加到我的 dplyr pipe 一个函数来转换上面的 dataframe变成类似下面的东西

Answer 1

假设在您的原始数据框中，在 category company_name 的第一个值的列在每一组中都标志着一个新组的开始，你可以这样做:

library(dplyr)
library(tidyr)

comp_df %>%
  group_by(category, grp = cumsum(category == "company_name" & lag(category, default = "") != "company_name")) %>%
  summarise(desc = paste(desc, collapse =  " ")) %>%
  pivot_wider(id_cols = grp, names_from = category, values_from = desc)

# A tibble: 3 x 6
    grp address              company_name    contact_person no_workers product   
  <int> <chr>                <chr>           <chr>          <chr>      <chr>     
1     1 12 West Road         AAA Company Ltd Bob C          344        fish      
2     2 North Perak          BBB Enteprises  Simon T        12         vegetables
3     3 Green Lane Singapore EF Industries   Sylvia P       8800       cement