Passing multiple dataframes into function to produce dataframe using map_dfr

I have a function which I am passing three variables (column names) and a dataframe and putting the results into a dataframe with map_dfr which works fine.

library(tidyverse)

iris_median<-function(df, col){
    df%>%
        summarise(Median=median(.data[[col]], na.rm=TRUE))%>%
        mutate(field=as.character(!!col))
}

median_df <- map_dfr(c("Sepal.Length","Sepal.Width",
                       "Petal.Length"), .f=iris_median,
                     df= iris)

When I try to pass multiple dataframes into this function as below:

iris1 <- iris [1:50,]
iris2 <- iris [51:100,]

median_df <- map_dfr("Sepal.Length", .f=iris_median,
                     df= c(iris1, iris2))

I get the error message: no applicable method for 'summarise' applied to an object of class "list"

I don't understand why it isn't just iterating through the two different dataframes and doing the calculation for each dataframe.

Can anyone ammend the code?

map_dfr iterates through the items in the first argument (the .x argument). It will not iterate through the additional arguments passed via ....

Since the map_ functions are just wrappers for a loop, your call

map_dfr("Sepal.Length", .f = iris_median, df = c(iris1, iris2))

calls the iris_median function once (since the first argument has length 1), and passes the whole list to the df argument at once. This is equivalent to

iris_median(df = c(iris1, iris2), col = "Sepal.Length")

And it is this call that causes your error, since you are trying to summarize a list.

To get map_dfr to work on a list of data frames, you need to pass the list as the first (.x) argument, and the single column name as an additional argument:

map_dfr(.x = list(iris1, iris2),
        .f = iris_median, 
        col = "Sepal.Length")
#>   Median        field
#> 1    5.0 Sepal.Length
#> 2    5.9 Sepal.Length

If you want to iterate through both a list of data frames and a vector of column names you need map2_dfr, which iterates through the first two arguments.

map2_dfr(.x = list(iris1, iris2), 
         .y = c("Sepal.Length", "Petal.Width"), 
         .f = iris_median)
#>   Median        field
#> 1    5.0 Sepal.Length
#> 2    1.3  Petal.Width

Update

Note that map2_dfr iterates through the .x and .y arguments in parallel. If you want all possible combinations of the data frames and column names, you would need to pass the data frames and column names through cross first.

df_list <- list(iris1, iris2)
col_vec <- c("Sepal.Length", "Sepal.Width")

cross(.l = list(df_list, col_vec)) %>%
  map_dfr(~ iris_median(.x[[1]], .x[[2]]))
#>   Median        field
#> 1    5.0 Sepal.Length
#> 2    5.9 Sepal.Length
#> 3    3.4  Sepal.Width
#> 4    2.8  Sepal.Width