Search code examples
rgroup-bydplyrmedian

R - Error computing interpolated mean by group

I am trying to compute the interpolated median by group for a number of variables. My dataframe looks like this:

# A tibble: 6 x 8
  id            eu_image eu_insurance eurobonds free_movement_welfare eu_cn_solidarity country_code country_party_mass
  <chr>            <dbl>        <dbl>     <dbl>                 <dbl>            <dbl> <dbl+lbl>    <chr>             
1 CAWI200000100        4            4         4                     3                3 2            germany_7         
2 CAWI300000784        2            2         1                     1                1 3            italy_9           
3 CAWI100000787        3            3         2                     2                3 1            france_13         
4 CAWI500000081        3            2         2                     1                3 5            spain_2           
5 CATI500000067        4            3         2                     2                6 5            spain_3           
6 CAWI100000398        2            4         4                     2                5 1            france_2

When I run the following code to compute the interpolated mean by the grouping variable country_party_mass:

party_median <- newdata %>%
    group_by(country_party_mass) %>%
    dplyr::summarise_at(c(   "eu_image", 
                      "eu_cn_solidarity", 
                      "eurobonds", 
                      "free_movement_welfare", 
                      "eu_insurance"), 
    funs(interp.median(., na.rm=TRUE))) %>%
    as.data.frame()

I get the following error:

Error in summarise_impl(.data, dots) : Column eu_cn_solidarity must be length 1 (a summary value), not 0

I have checked previous questions on similar issues, but I could not find a viable solution.

Solution

  • Building on A. Suliman's comment: you can add an ifelse function to check if all entries are NA:

    party_median <- newdata %>%
    group_by(country_party_mass) %>%
    dplyr::summarise_at(vars(c("eu_image", 
                      "eu_cn_solidarity", 
                      "eurobonds", 
                      "free_movement_welfare", 
                      "eu_insurance")), 
    ~ifelse(all(is.na(.)), NA_real_, interp.median(., na.rm=TRUE)))

    Note that the funs function is now soft deprecated (as of dplyr 0.8.0.1) so I use the "~" notation instead. Also I use the vars function to select variables.