如何在 R 中的两个多类别数据列 table 之间找到 difference/setdiff()

Question

我的数据在两个数据表中，如下所示（比此处显示的列多得多） -

数据表 1 = data_sale

Site Id	Country	Product ID
1000375476	Canada	UG10000-WISD
1000375476	Canada	UGD12895
1000706152	Switzerland	UG10000-WISD
1000706152	Switzerland	UG80000-NTCD-G
1000797366	Italy	UG10000-WISD
1000797366	Italy	UG12210

数据表 2 = data_licenses

Site Id	Country	Product ID
1000375476	Canada	UG10000-WISD
1000375476	Canada	UGD12895
1000797366	Italy	UG12785
1000797366	Italy	UG12210

I want to calculate the set difference for unique Product ID for all the Site Id in data_sale, keeping all rows.

这是我到目前为止所做的 -

对于这两个数据表，我创建了一个新列，其中包含所有独特的产品。

data_sale <-
  data_sale[, `unique_products` := paste0(unique(`Product ID`), collapse = ","), 
              keyby = c("Site Id")]

data_licenses <-
  data_licenses[, .(`unique_products` = paste0(unique(`Product ID`), collapse = ",")), 
              keyby = c("Site Id")]

左合并 data_sale 与 data_licenses

merge(data_sale, data_licenses, by = 'Site Id', all.x = TRUE)

现在合并后的数据表是这样的 -

Site Id	Country	Product ID	unique_products.data_sale	unique_products.data_licenses
1000375476	Canada	UG10000-WISD	UG10000-WISD,UGD12895	UG10000-WISD,UGD12895
1000375476	Canada	UGD12895	UG10000-WISD,UGD12895	UG10000-WISD,UGD12895
1000706152	Switzerland	UG10000-WISD	UG10000-WISD,UG80000-NTCD-G	NA
1000706152	Switzerland	UG80000-NTCD-G	UG10000-WISD,UG80000-NTCD-G	NA
1000797366	Italy	UG10000-WISD	UG10000-WISD,UG12210	UG12785,UG12210
1000797366	Italy	UG12210	UG10000-WISD,UG12210	UG12785,UG12210

The problem is with my final step where I want a new column showing difference between the products of data_sale and data_licenses, it should look like this -

Site Id	Country	Product ID	unique_products.data_sale	unique_products.data_licenses	difference
1000375476	Canada	UG10000-WISD	UG10000-WISD,UGD12895	UG10000-WISD,UGD12895	NA
1000375476	Canada	UGD12895	UG10000-WISD,UGD12895	UG10000-WISD,UGD12895	NA
1000706152	Switzerland	UG10000-WISD	UG10000-WISD,UG80000-NTCD-G	NA	UG10000-WISD,UG80000-NTCD-G
1000706152	Switzerland	UG80000-NTCD-G	UG10000-WISD,UG80000-NTCD-G	NA	UG10000-WISD,UG80000-NTCD-G
1000797366	Italy	UG10000-WISD	UG10000-WISD,UG12210	UG12785,UG12210	UG10000-WISD
1000797366	Italy	UG12210	UG10000-WISD,UG12210	UG12785,UG12210	UG10000-WISD

关于如何实现它的任何线索都会有很大帮助。谢谢！

下面是合并数据表使用dput()得到的数据

structure(list(`Site Id` = c("1000375476", "1000375476", "1000706152", 
"1000706152", "1000797366", "1000797366"), Country = c("Canada", 
"Canada", "Switzerland", "Switzerland", "Italy", "Italy"), `Product ID` = c("UG10000-WISD", 
"UGD12895", "UG10000-WISD", "UG80000-NTCD-G", "UG10000-WISD", 
"UG12210"), unique_products.x = c("UG10000-WISD,UGD12895", "UG10000-WISD,UGD12895", 
"UG10000-WISD,UG80000-NTCD-G", "UG10000-WISD,UG80000-NTCD-G", 
"UG10000-WISD,UG12210", "UG10000-WISD,UG12210"), unique_products.y = c("UG10000-WISD,UGD12895", 
"UG10000-WISD,UGD12895", NA, NA, "UG12785,UG12210", "UG12785,UG12210"
)), sorted = "Site Id", class = c("data.table", "data.frame"), row.names = c(NA, 
-6L), .internal.selfref = <pointer: 0x556bb5c10a40>)

Answer 1

可能有一种方法可以尝试组合一些 built-in 函数，但是有一个简单的自定义函数的例子：

find_differences = function(x,y){
  # x: column list of strings we want to compare to
  # y: other column list
  x = strsplit(x,',') # transform strings to lists
  y = strsplit(y,',')
  
  differences = list()

  for(i in seq(1,length(x))){  # for every row (nested-list)
    
    if(identical(x[[i]],y[[i]])){
      row_diff = NA
    }
    else{
    row_diff = paste(x[[i]][ ! x[[i]] %in% y[[i]] ],collapse=',')
    }
    
    differences = c(differences,row_diff)
  }
  return(differences)
}

以你为例：

example = rename(example, 
                 unique_products.data_sale = unique_products.x,
                 unique_products.data_licenses = unique_products.y)

example$difference = find_differences(example$unique_products.data_sale, example$unique_products.data_license)

> example
      Site Id     Country     Product ID   unique_products.data_sale unique_products.data_licenses                  difference
1: 1000375476      Canada   UG10000-WISD       UG10000-WISD,UGD12895         UG10000-WISD,UGD12895                          NA
2: 1000375476      Canada       UGD12895       UG10000-WISD,UGD12895         UG10000-WISD,UGD12895                          NA
3: 1000706152 Switzerland   UG10000-WISD UG10000-WISD,UG80000-NTCD-G                          <NA> UG10000-WISD,UG80000-NTCD-G
4: 1000706152 Switzerland UG80000-NTCD-G UG10000-WISD,UG80000-NTCD-G                          <NA> UG10000-WISD,UG80000-NTCD-G
5: 1000797366       Italy   UG10000-WISD        UG10000-WISD,UG12210               UG12785,UG12210                UG10000-WISD
6: 1000797366       Italy        UG12210        UG10000-WISD,UG12210               UG12785,UG12210                UG10000-WISD

Answer 2

这将得到 data_sale 中 data_license 中没有的产品 Site Id。与其连接唯一产品 ID，不如将唯一列用作字符向量更容易。

library(data.table)

data_licenses <- data.table(`Site Id` = c("1000375476", "1000375476", "1000797366", "1000797366"),
                            Country = c("Canada", "Canada", "Italy", "Italy"),
                            `Product ID` = c("UG10000-WISD", "UGD12895", "UG12785", "UG12210"))
data_sale <- data.table(`Site Id` = c("1000375476", "1000375476", "1000706152", "1000706152", "1000797366", "1000797366"),
                        Country = c("Canada", "Canada", "Switzerland", "Switzerland", "Italy", "Italy"),
                        `Product ID` = c("UG10000-WISD", "UGD12895", "UG10000-WISD", "UG80000-NTCD-G", "UG10000-WISD", "UG12210"))

data_unique <- data_sale[
  , .(unique_products.data_sale = .(unique(`Product ID`))), c("Site Id", "Country")
][
  data_licenses[, .(unique_products = .(unique(`Product ID`))), "Site Id"],
  unique_products.data_licenses := i.unique_products,
  on = "Site Id"
][
  , difference := lapply(.I, function(i) setdiff(unique_products.data_sale[[i]], unique_products.data_licenses[[i]]))
]
print(data_unique)
#>       Site Id     Country   unique_products.data_sale unique_products.data_licenses                  difference
#> 1: 1000375476      Canada       UG10000-WISD,UGD12895         UG10000-WISD,UGD12895                            
#> 2: 1000706152 Switzerland UG10000-WISD,UG80000-NTCD-G                               UG10000-WISD,UG80000-NTCD-G
#> 3: 1000797366       Italy        UG10000-WISD,UG12210               UG12785,UG12210                UG10000-WISD

Answer 3

这种快速获取每个站点差异的方法怎么样？然后您可以将结果合并回具有 Site Id:

的任何帧

data_licenses[, .(licen_p = .(.(`Product ID`))), by = `Site Id`] %>% 
  .[data_sale[, .(sale_p= .(.(`Product ID`))), by=`Site Id`],on=.(`Site Id`)] %>% 
  .[,.(difference = toString(unlist(setdiff(sale_p, licen_p)))), by=`Site Id`]

输出：

      Site Id                   difference
1: 1000375476                             
2: 1000706152 UG10000-WISD, UG80000-NTCD-G
3: 1000797366        UG10000-WISD, UG12210

如何在 R 中的两个多类别数据列 table 之间找到 difference/setdiff()

How to find difference/setdiff() between two multi-category columns of data table in R

datatable

group-by

r

set-difference

data.table