Error in dplyr::select() while writing a function-CodePudding

I am writing a function to conduct one way anova and was writing the below function

fun_aov <- function(sps, param){
  
  ths_data_aov <-  ths_data_aov |> filter(Species == sps) |> 
    select(Species, Treatment, param)
  
  n_one_way_aov <- aov( param ~ Treatment, data = ths_data_aov)
  h <- summary(n_one_way_aov)
  
  return(h)
}

for the data frame below

Species Treatment    num_roots_n lng_long_root_cm dia_long_root_mm
   <chr>   <chr>           <dbl>          <dbl>        <dbl>
  x1       t1                  4            7          0.6
  x1       t1                  4            7          0.6
  x1       t1                  4            7          0.6
  x1       t1                  4            7          0.6
  x1       t2                  4            8          0.7
  x1       t2                  3            6          0.8
  x1       t2                  4            8          0.9
  x1       t2                  5            7          0.3
  x1       t3                  8            8          0.5
  x1       t3                  3            5          0.7
  x1       t3                  4            6          0.3
  x1       t3                  5            5          0.7
  x1       t4                  6            4          0.7
  x1       t4                  9            3          0.8
  x1       t4                  9            3          0.8
  x1       t4                  9            3          0.8

but when I execute the function fun_aov("x1", lng_long_root_cm) an error shows up saying Error in `select()`: ! object 'lng_long_root_cm' not found how do I rectify it.

I am expecting the return of h which gives me an analysis wrt to particular Species and param

CodePudding user response：

The difficulty here is that your function uses two different functions which themselves rely on two different types of non-standard evaluation. If you use a dplyr verb such as select inside a function, and wish to pass an unquoted column name as an argument (such as param in your example), then you need to use the curly-curly operator, i.e. {{param}} inside select, otherwise it will look for an actual column called param inside your data frame, which of course doesn't exist.

Similarly, variable names inside a formula are not substituted for parameters passed into your function, so aov will be looking for a column called param inside your data frame. Here, you need to construct the formula with something like as.formula and pass it with do.call to get this working correctly:

library(dplyr)

fun_aov <- function(sps, param) {
  
  ths_data_aov <-  ths_data_aov %>% 
    filter(Species == sps) %>% 
    select(Species, Treatment, {{param}})
  
  aov_f <- as.formula(paste(deparse(substitute(param)), '~ Treatment'))
  n_one_way_aov <- do.call('aov', list(formula = aov_f, data  = ths_data_aov))
  
  summary(n_one_way_aov)
}

Now you have

fun_aov("x1", lng_long_root_cm)
#>             Df Sum Sq Mean Sq F value  Pr(>F)    
#> Treatment    3  40.25  13.417   16.95 0.00013 ***
#> Residuals   12   9.50   0.792                    
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Data in reproducible format

ths_data_aov <- structure(list(Species = c("x1", "x1", "x1", "x1", "x1", "x1", 
"x1", "x1", "x1", "x1", "x1", "x1", "x1", "x1", "x1", "x1"), 
    Treatment = c("t1", "t1", "t1", "t1", "t2", "t2", "t2", "t2", 
    "t3", "t3", "t3", "t3", "t4", "t4", "t4", "t4"), num_roots_n = c(4L, 
    4L, 4L, 4L, 4L, 3L, 4L, 5L, 8L, 3L, 4L, 5L, 6L, 9L, 9L, 9L
    ), lng_long_root_cm = c(7L, 7L, 7L, 7L, 8L, 6L, 8L, 7L, 8L, 
    5L, 6L, 5L, 4L, 3L, 3L, 3L), dia_long_root_mm = c(0.6, 0.6, 
    0.6, 0.6, 0.7, 0.8, 0.9, 0.3, 0.5, 0.7, 0.3, 0.7, 0.7, 0.8, 
    0.8, 0.8)), class = "data.frame", row.names = c(NA, -16L))

CodePudding user response：

When you use a variable on r function you need to specify in quote in the function call. You also need to add select(all_of(param)) to select all the values in the :

fun_aov <- function(sps, param){
  
  ths_data_aov <-  ths_data_aov |> filter(Species == sps) |> 
    select(Species, Treatment, all_of(param))
  
  n_one_way_aov <- aov( param ~ Treatment, data = ths_data_aov)
  h <- summary(n_one_way_aov)
  
  return(h)
}

fun_aov("x1", 'lng_long_root_cm')