I am trying to use pivot_wider inside a function. However, I cannot access the new columns once I have changed the format. I get the error message that columns a and b don't exist.

This is my function:

dip <- function(df, a, b, year_min) {
  
  wow <- df %>%  
  filter(country %in% c("a","b")) %>% 
  filter(date >= year_min) %>% 
  group_by(rcid) %>% 
  pivot_wider(id_cols = "rcid", 
              names_from = c("a","b"), 
              values_from = "vote") %>% 
  na.omit()
  
  percentage <- sum(wow$a == wow$b)*100/nrow(wow_df)
  
  return(percentage)
}

dip(joined_df, "United States", "Germany", 1990-01-01)

When I try to run the code without the function, it works:

joined_df <- un_votes %>%
  inner_join(un_roll_calls, by = "rcid")
   
wow_df <- joined_df %>% 
  filter(country == "United States" | country == "Germany") %>% 
  filter(date >= "1990-01-01") %>% 
  group_by(rcid) %>% 
  pivot_wider(id_cols = "rcid", 
              names_from = country, 
              values_from = "vote") %>% 
  na.omit()

percentage <- sum(wow_df$"Germany" == wow_df$"United States")*100/nrow(wow_df)

I am fairly new to function and appreciate any tips. The data I'm using is from the un_votes packages.

This is the error message I get:

Error in chr_as_locations(): ! Can't subset columns that don't exist. ✖ Column a doesn't exist. Backtrace:

1. global dip(joined_df, "United States", "Germany", 1990 - 1 - 1)
2. tidyr:::pivot_wider.data.frame(...)
3. tidyr::build_wider_spec(...)
4. tidyselect::eval_select(enquo(names_from), data)
5. tidyselect:::eval_select_impl(...) ...
6. tidyselect:::reduce_sels(node, data_mask, context_mask, init = init)
7. tidyselect:::walk_data_tree(new, data_mask, context_mask)
8. tidyselect:::as_indices_sel_impl(...)
9. tidyselect:::as_indices_impl(x, vars, call = call, strict = strict)
10. tidyselect:::chr_as_locations(x, vars, call = call) Error in chr_as_locations(x, vars, call = call)

CodePudding user response：

Here we need to unquote the "a", "b". In addition, replace the $ with [[

dip <- function(df, a, b, year_min) {
  
  wow <- df %>%  
  filter(country %in% c(a, b)) %>% 
  filter(date >= as.Date(year_min)) %>%   
  pivot_wider(id_cols = "rcid", 
              names_from = country, 
              values_from = "vote") %>% 
  na.omit()
  
  percentage <- sum(wow[[a]] == wow[[b]])*100/nrow(wow)
  
  return(percentage)
}

-testing

dip(joined_df, "United States", "Germany", "1990-01-01")
[1] 33.33333

data

joined_df <- structure(list(rcid = c(1L, 2L, 3L, 4L, 5L, 1L, 2L, 3L, 4L, 5L, 
1L, 2L, 3L, 4L, 5L), country = c("United States", "United States", 
"United States", "United States", "United States", "Germany", 
"Germany", "Germany", "Germany", "Germany", "Spain", "Spain", 
"Spain", "Spain", "Spain"), date = structure(c(12904, 8066, -25932, 
8401, 12843, 11231, 6971, 10470, 9251, 13787, 14304, 14396, 13361, 
11566, 8126), class = "Date"), vote = c(73, 88, 25, 73, 76, 73, 
91, 25, 88, 31, 45, 34, 80, 66, 60)), row.names = c(NA, -15L),
 class = "data.frame")