CODEWITHSUNDEEP
rdataframe
Djoustaine
Djoustaine

Reputation: 95

Add rows and complete dyad by group

I have a dataset in a dyadic format and sorted by group and I am trying to add an observation to each group. I need this observation to also be integrated with the other pairs. Below is a reproducible example to show what I mean. Data is a simplified version of my dataset (it contains more groups essentially).

data <- data.frame(country1 = c("BEL", "FRA", "BEL", "FRA", "AUS", "ITA"),
                   country2 = c("FRA", "BEL", "FRA", "BEL", "ITA", "AUS"),
                   year = c(2001,2001,2002,2002,2002,2002),
                   id = c(1,1,1,1,2,2))
> data
   country1 country2 year id
1       BEL      FRA 2001  1
2       FRA      BEL 2001  1
3       BEL      FRA 2002  1
4       FRA      BEL 2002  1
5       AUS      ITA 2002  2
6       ITA      AUS 2002  2

I would like to add a different country to each group. For instance, say I would like to add Luxembourg to group 1 and Portugal to group 2.

This is what the output I need should look like:

> data
   country1 country2 year id
1       BEL      FRA 2001  1
2       FRA      BEL 2001  1
3       LUX      BEL 2001  1
4       LUX      FRA 2001  1
5       BEL      LUX 2001  1
6       FRA      LUX 2001  1
7       BEL      FRA 2002  1
8       FRA      BEL 2002  1
9       LUX      BEL 2002  1
10      LUX      FRA 2002  1
11      BEL      LUX 2002  1
12      FRA      LUX 2002  1
13      AUS      ITA 2002  2
14      ITA      AUS 2002  2
15      POR      AUS 2002  2
16      POR      ITA 2002  2
17      AUS      POR 2002  2
18      ITA      POR 2002  2

I found a workaround way but I don't know how to simplify this process and to automate it to some extent.

id1 <- data%>%
  filter(id== 1) %>%
  mutate(country3 = "LUX")
id1_1 <- id1 %>%
  select(!country2) %>%
  rename("country2" = "country3") %>%
  distinct()
id1_2 <- id1 %>%
  select(!country1) %>%
  rename("country1" = "country3") %>%
  distinct()
id1_2 <- id1_2 [, c(2,1,3,4)]
id1 <- rbind(id1_1, id1_2)
data<- rbind(data, id1)

This completes the dyads but it is quite tedious to do since I am trying to add about 100 countries to a hundred groups.

I can create either a vector or a data frame containing all the countries I need to add (and arrange them by group if necessary), but I just don't know how to use them to fill the main data. Thanks for any tips!

Upvotes: 1

Views: 79

Answers (1)

Carl
Carl

Reputation: 7540

Would something like this work for you?

library(tidyverse)

data <- data.frame(country1 = c("BEL", "FRA", "BEL", "FRA", "AUS", "ITA"),
                   country2 = c("FRA", "BEL", "FRA", "BEL", "ITA", "AUS"),
                   year = c(2001,2001,2002,2002,2002,2002),
                   id = c(1,1,1,1,2,2))

additions <- tribble(
  ~id, ~country1,
  1, "LUX",
  2, "POR"
)

unique_combos <- data |> 
  distinct(id, year, country1) |> 
  rows_append(additions) |> 
  expand(year, nesting(id, country1)) |> 
  filter(!is.na(year))

unique_combos |> 
  rename(country2 = country1) |> 
  full_join(unique_combos) |> 
  filter(country1 != country2) |> 
  arrange(id, year, country1, country2)

#> Joining, by = c("year", "id")
#> # A tibble: 24 × 4
#>     year    id country2 country1
#>    <dbl> <dbl> <chr>    <chr>   
#>  1  2001     1 FRA      BEL     
#>  2  2001     1 LUX      BEL     
#>  3  2001     1 BEL      FRA     
#>  4  2001     1 LUX      FRA     
#>  5  2001     1 BEL      LUX     
#>  6  2001     1 FRA      LUX     
#>  7  2002     1 FRA      BEL     
#>  8  2002     1 LUX      BEL     
#>  9  2002     1 BEL      FRA     
#> 10  2002     1 LUX      FRA     
#> # … with 14 more rows

Created on 2022-06-29 by the reprex package (v2.0.1)

Upvotes: 1

Related Questions