How to use for loop to extract rows with similar elements across 2 dataframes? - r

Question

I have 2 dataframes, one is a Free Trade Agreement dataset that contains many columns, the columns c1 to c91 denote different countries part of a particular Free Trade Agreement, as shown below:
FTA data
FTA data e.g.
No Base_treaty entry_type c1 c2 c3
1 1 treaty Afghanistan India NA
2 2 treaty Algeria Egypt Ghana
3 3 treaty Algeria Angola Benin
4 4 treaty Egypt Jordan Morocco
5 5 treaty Albania Bulgaria NA
6 6 treaty Albania Croatia NA
The other data frame contains trade data between two particular countries, i and j. Trade Data
inventor_ctry_i authority_ctry_j
1 Albania Bulgaria
2 Albania Croatia
3 Algeria Angola
4 Algeria Belgium
5 Algeria France
6 Andorra Turkey
7 Andorra United States
8 Anguilla Germany
9 Anguilla Switzerland
10 Anguilla United States
Desired output:
No Base_treaty entry_type matched ctry1 matched ctry2
3 3 treaty Algeria Angola
5 5 treaty Albania Bulgaria
6 6 treaty Albania Croatia
I want to be able to find countries i and j in trade data that show up in the same row somewhere in between c1 to c91 of the FTA data. If both are present in a particular row, extract the 2 countries from the row in FTA, keeping no, base treaty and entry type column intact.
What I have done so far:
FTA_final: FTA Data, unique_pairs: Trade Data
specialnames <- setdiff(names(FTA_final), c("number", "base_treaty",
"entry_type")) **#getting rid of irrelevant columns**
table <- data.frame()` **#create empty dataframe**
for(i in nrow(FTA_final)){`
for(j in seq_along(specialnames)){`
for(p in nrow(unique_pairs)){`
if (FTA_final[i,j] %in% unique_pairs[p,])`
{table <- rbind(table,FTA_final[i,c(1:3, j)])}`
` }`
`}`
`}` **#for loop**
Nothing happens when I run these codes, not sure why. Any help would be greatly appreciated.

Answer 1

One way to do this would be to row-wise paste the value of Trade_data to get combinations of countries that trade together. We can then create a combination of countries in FTA_data and check if any of the combination matches all_countries.
cols <- paste0('c', 1:3)
all_countries <- do.call(paste, Trade_data)
data <- apply(FTA_data[cols], 1, function(x) {
x <- na.omit(x)
if(length(x) <= 1) return(NULL)
temp <- combn(x, 2)
inds <- combn(x, 2, paste, collapse = " ") %in% all_countries
if(any(inds)) temp[, inds]
})
new_data <- FTA_data[!sapply(data, is.null), ]
new_data[cols] <- NULL
final_data <- cbind(new_data, do.call(rbind, data))
final_data
# No Base_treaty entry_type 1 2
#3 3 3 treaty Algeria Angola
#5 5 5 treaty Albania Bulgaria
#6 6 6 treaty Albania Croatia
Here is another way :
library(dplyr)
library(tidyr)
output<- FTA_data[rowSums(sapply(all_countries, function(x)
apply(FTA_data[cols], 1, function(y)
grepl(x, paste(y, collapse = " "))))) > 0, ]
output %>%
pivot_longer(cols = starts_with('c'),
values_drop_na = TRUE) %>%
filter(value %in% Trade_data$inventor_ctry_i |
value %in% Trade_data$authority_ctry_j) %>%
group_by(No, Base_treaty, entry_type) %>%
mutate(name = paste0('c', row_number())) %>%
pivot_wider()

Answer 2

Thank you to #Ronak Shah for your suggestions
As suggested by #Ronak Shah, I was able to get the relevant rows that had countries i and j in them:
cols <- paste0('c', 1:3)
all_countries <- do.call(paste, Trade_data)
output<- FTA_data[rowSums(sapply(all_countries, function(x)
apply(FTA_data[cols], 1, function(y)
grepl(x, paste(y, collapse = " "))))) > 0, ]
Afterwhich, I did this:
do.call(rbind, combn(grep("^c\\d+$", names(output)), 2, function(x)
cbind(output[1:3], setNames(output[x], paste0("c", 1:2))), simplify=F))
This helps me get all possible combinations across "c" columns, while retaining columns 1:3, i.e. No, Base Entry and entry_type.
After this, a simple left join with trade data will gave me the desired i and j pairs and the output:
No Base_treaty entry_type matched ctry1 matched ctry2
3 3 treaty Algeria Angola
5 5 treaty Albania Bulgaria
6 6 treaty Albania Croatia