Replacing exact string with a word from a dataframe & matching strings just containing a certain word

user_14015186 0
Viewed 230

I am working with R and I have two data frames. One data frame my_data is my main dataset that contains order data, the other one, word_list, contains a list of words that I would like to match with my_data.

Here is a reproducible example of the two data frames:

my_data <- data.frame(
  Order = c("1","2", "3", "4", "5", "6"),
  Product_ID = c("TS678", "AB123", "PACK12, 1xGF123, 1xML680", "AB123", "PACK13, 1xML680, 1x2304TR", "GF123"))

word_list <- data.frame(
  Codes = c("TS678","AB123", "GF123", "CC756"),
  Product_Category = c("Apple", "Apple", "Orange", "Orange"))

What I would like to do is to match the Product_ID in my_data with the Codes in word_list and add a new column to my_data with the matching Product_Category from word_list.
However, I need to implement exact matches as well as consider Code combinations (as seen with "PACK" in the sample data, which consists of multiple product codes in one column)

For the final dataframe I want to end up with the following:

  1. Match the exact matches -> add the corresponding Product_Category, e.g. "Apple"
  2. Match columns that contain the code from word_list, but also contain other codes. Certain products are Packs and the ID is mixed with other IDs -> this should result "Apple + Other" if the code for "apple" is contained plus other codes are contained. Another issue here is that the Code that needs to be matched is also accompanied by a count (e.g., PACK12 includes 1x GF123, 1xML680, etc.)
  3. All columns that do not contain the exact match nor a mixed match should be assigned "Other"

To make it better understandable, what I would like to get as a final result is a dataframe that looks like the following:

 my_data_result <- data.frame(
  Order = c("1","2", "3", "4", "5", "6"),
  Product_ID = c("TS678", "AB123", "PACK12, 1xGF123, 1xML680", "AB123", "PACK13, 1xML680, 1x2304TR", "GF123"),
  Product_Category = c("Apple", "Apple", "Orange + Other", "Apple", "Other", "Orange"))

I assume this could be done with regex & gsub, but I am not sure how.

Thank you!

gsub dataframe r text
2 Answers

Since your data is huge you can try this data.table approach :

library(data.table)
library(splitstackshape)

#Convert to data.table
setDT(my_data)
setDT(word_list)

#Get the data in long format
df1 <- cSplit(my_data, 'Product_ID', direction = 'long')
#Remove initial characters 
df1[, Product_ID := sub('.*x', '', Product_ID)]

#Join the dataframes
df1 <- merge(df1, word_list, by.x = 'Product_ID', by.y = 'Codes', all.x = TRUE)
#Replace NA with "Other"
df1[, Product_Category := replace(Product_Category, 
                           is.na(Product_Category), 'Other')]

#Combine the values by Order
df1[, .(Product_ID = toString(Product_ID), 
       Product_Category = paste(sort(unique(Product_Category)), 
                          collapse = " + ")), Order]

#   Order            Product_ID Product_Category
#1:     5 2304TR, ML680, PACK13            Other
#2:     2                 AB123            Apple
#3:     4                 AB123            Apple
#4:     3  GF123, ML680, PACK12   Orange + Other
#5:     6                 GF123           Orange
#6:     1                 TS678            Apple
user_39629140

Here is an idea using dplyr and tidyr. We split the rows to long, clean the codes, match with word_list and convert back to strings per Order, i.e.

library(dplyr)
library(tidyr)

my_data %>% 
 separate_rows(Product_ID, sep = ', ') %>% 
 mutate(Product_ID = sub('.*x', '', Product_ID), 
        Product_Category = as.character(word_list$Product_Category[match(Product_ID, word_list$Codes)]), 
        Product_Category = replace(Product_Category, is.na(Product_Category), 'Other')) %>%
 group_by(Order) %>% 
 summarise_all(list(~toString(unique(.))))

# A tibble: 6 x 3
#  Order Product_ID            Product_Category
#  <fct> <chr>                 <chr>           
#1 1     TS678                 Apple           
#2 2     AB123                 Apple           
#3 3     PACK12, GF123, ML680  Other, Orange   
#4 4     AB123                 Apple           
#5 5     PACK13, ML680, 2304TR Other           
#6 6     GF123                 Orange
user_56355800
Related