R - Problem Converting Multiple Columns to Numeric-CodePudding

I have several dataframes in my dataset that contain numeric values, but are type chr. They look like this:

   IDs     FCC_Faces FCNC_Faces          FCC_Objects         FCNC_Objects  FCC_Scenes FCNC_Scenes FCC_Words  FCNC_Words    Age Group
   <chr>       <dbl> <chr>               <chr>               <chr>         <chr>      <chr>       <chr>      <chr>       <dbl> <dbl>
 1 CON_L01      1    0.58330000000000004 0.83330000000000004 0.5833000000~ 0.6        0.6         0.9165999~ 0.66659999~    63     0
 2 CON_L03      0.92 0.83                1                   0.92          0.9        0.9         0.83       1              37     0
 3 CON_L04      0.75 0.83                0.92                0.92          0.9        0.8         0.67       0.92           48     0
 4 CON_L05      0.75 1                   0.92                1             0.9        1           1          1              49     0
 5 CON_L07      0.58 0.17                0.57999999999999996 0.75          0.8        0.7         0.83       0.67           69     0
 6 CON_L10      0.92 0.67                0.83                0.5799999999~ 0.8        0.9         0.33       0.57999999~    58     0
 7 CON_L14      0.83 0.83                0.83                0.75          0.8        0.9         0.92       0.67           62     0
 8 CON_L16      1    0.92                NA                  NA            0.9        0.9         1          1              40     0
 9 CON_L17      0.83 0.57999999999999996 0.75                0.83          0.9        0.6         1          0.83           48     0
10 CON_L18      0.75 0.75                0.75                0.75          0.9        0.5         0.75       0.67           55     0
# ... with 70 more rows

I wanted to write a function to which I could pass the dataframe, the column names, and have it convert all of them to numeric. My first attempt was just using lapply():

cols_to_numeric <- function(dframe, columns) {
    dframe[ , columns] <- lapply(dframe[ , columns], numeric)
    return(dframe)
}

However, I get this error:

Error in FUN(X[[i]], ...) : invalid 'length' argument 
3.
FUN(X[[i]], ...) 
2.
lapply(dframe[, columns], numeric)

So, looking for another solution, this thread suggested a nice approach using Tidyverse:

cols_to_numeric <- function(dframe, columns) {
    require(magrittr)
    require(tidyverse)
    
    dframe %<>% mutate_at(columns, numeric)
    return(dframe)
}

But I get a very similar error:

 Error: Problem with `mutate()` input `FCC_Faces`.
x invalid 'length' argument
i Input `FCC_Faces` is `(function (length = 0L) ...`.

But the length of this variable is very clearly not zero:

> length(CON_data$FCC_Faces)
[1] 80

Further, if I just do the coercion for each column manually, it works without complaints:

> CON_data$FCC_Faces <- as.numeric(CON_data$FCC_Faces)

> str(CON_data)
'data.frame':   80 obs. of  11 variables:
 $ IDs         : chr  "CON_L01" "CON_L03" "CON_L04" "CON_L05" ...
 $ FCC_Faces   : num  1 0.92 0.75 0.75 0.58 0.92 0.83 1 0.83 0.75 ...
 $ FCNC_Faces  : chr  "0.58330000000000004" "0.83" "0.83" "1" ...
 $ FCC_Objects : chr  "0.83330000000000004" "1" "0.92" "0.92" ...
 $ FCNC_Objects: chr  "0.58330000000000004" "0.92" "0.92" "1" ...
 $ FCC_Scenes  : chr  "0.6" "0.9" "0.9" "0.9" ...
 $ FCNC_Scenes : chr  "0.6" "0.9" "0.8" "1" ...
 $ FCC_Words   : chr  "0.91659999999999997" "0.83" "0.67" "1" ...
 $ FCNC_Words  : chr  "0.66659999999999997" "1" "0.92" "1" ...
 $ Age         : num  63 37 48 49 69 58 62 40 48 55 ...
 $ Group       : num  0 0 0 0 0 0 0 0 0 0 ...
>

What on earth am I doing wrong here? I don't want to have to manually coerce every single column for each dataframe every time I make any changes.

CodePudding user response：

A possible solution:

library(dplyr)

df <- tibble(
  x = c("0.2", "0.4", "0.9"), 
  y = c("1.2", "2.4", "3.9"))

df %>% 
  mutate(across(x:y, as.numeric))
#> # A tibble: 3 × 2
#>       x     y
#>   <dbl> <dbl>
#> 1   0.2   1.2
#> 2   0.4   2.4
#> 3   0.9   3.9

CodePudding user response：

Just use type.convert function from base R:

type.convert(dframe, as.is = TRUE)