Loop through 2 dataframes to identify common columns - r

Question

I have 2 reproducible dataframes over here. I am trying to identify which column contain values that are similar to another column. I hope my code will take in every row and loop through every single column in df2. My code works below, but it requires fine-tuning to allow multiple matches with the same column.
df1 <- data.frame(fruit=c("Apple", "Orange", "Pear"), location = c("Japan", "China", "Nigeria"), price = c(32,53,12))
df2 <- data.frame(grocery = c("Durian", "Apple", "Watermelon"),
place=c("Korea", "Japan", "Malaysia"),
name = c("Mark", "John", "Tammy"),
favourite.food = c("Apple", "Wings", "Cakes"),
invoice = c("XD1", "XD2", "XD3"))
df <- sapply(names(df1), function(x) {
temp <- sapply(names(df2), function(y)
if(any(match(df1[[x]], df2[[y]], nomatch = FALSE))) y else NA)
ifelse(all(is.na(temp)), NA, temp[which.max(!is.na(temp))])
}
)
t1 <- data.frame(lapply(df, type.convert), stringsAsFactors=FALSE)
t1 <- data.frame(t(t1))
t1 <- cbind(newColName = rownames(t1), t1)
rownames(t1) <- 1:nrow(t1)
colnames(t1) <- c("Columns from df1", "Columns from df2")
df1
fruit location price
1 Apple Japan 32
2 Orange China 53
3 Pear Nigeria 12
df2
grocery place name favourite.food invoice
1 Durian Korea Mark Apple XD1
2 Apple Japan John Wings XD2
3 Watermelon Malaysia Tammy Cakes XD3
t1 #(OUTPUT FROM CODE ABOVE)
Columns from df1 Columns from df2
1 fruit grocery
2 location place
3 price <NA>
This is the output I hope to obtain instead:
Columns from df1 Columns from df2
1 fruit grocery, favourite.food
2 location place
3 price <NA>
Notice that the columns, "Grocery" and "favourite.food" both matches to the column "fruit", whereas my code only returns one column.

Answer 1

We can change the code to return all the matches instead and wrap them in one string using toString
vec <- sapply(names(df1), function(x) {
temp <- sapply(names(df2), function(y)
if(any(match(df1[[x]], df2[[y]], nomatch = FALSE))) y else NA)
ifelse(all(is.na(temp)), NA, toString(temp[!is.na(temp)]))
}
)
vec
# fruit location price
#"grocery, favourite.food" "place" NA
To convert it into dataframe, we can do
data.frame(columns_from_df1 = names(vec), columns_from_df2 = vec, row.names = NULL)
# columns_from_df1 columns_from_df2
#1 fruit grocery, favourite.food
#2 location place
#3 price <NA>