dplyr summarise function not working with global environment variables

Question

I'm trying to write a function that calculates the proportion of one column (outcome) given the values of another column. The code looks like this:

thresh_measure <- function(data, indicator, thresh_value)
{
   d1 <- data %>% 
    group_by(class_number, outcome) %>%
    summarize(n=sum(indicator <= thresh_value)) %>% spread(outcome, n)
    d1$thresh_value <- thresh_value
    return(d1)
}

final_test <- thresh_measure(df, 'pass_rate', 0.8)

There seems to be an error with the summarise function where the current function returns all 0's. When I change it to look like this, it works:

thresh_measure <- function(data, indicator, thresh_value)
{
   d1 <- data %>% 
    group_by(class_number, outcome) %>%
    summarize(n=sum(pass_rate <= thresh_value)) %>% spread(outcome, n)
    d1$thresh_value <- thresh_value
    return(d1)
}

final_test <- thresh_measure(df, 'pass_rate', 0.8)

I've tried using the .GlobalEnv to set the value, I've also detached all libraries except dplyr but it still isn't working.

Answer 1

You have to deal with the name of the column you want to pass as a parameter .. For exemple (certainly better ways exists) :

thresh_measure <- function(data, indicator, thresh_value)
{
  d1 <- data
 names(d1)[names(d1)==indicator] <- "indicator"
  d1 <- d1 %>% 
    group_by(class_number, outcome)  %>%
    summarize(n=sum(indicator <= thresh_value))  %>% spread(outcome, n)

   d1$thresh_value <- thresh_value
  return(d1)
}

Answer 2

Two alternative ways that should work:

# alternative I
thresh_measure <- function(data, indicator, thresh_value)
{
    ind_quo <- rlang::sym(indicator)
    d1 <- data %>%
        group_by(class_number, outcome) %>%
        summarize(n=sum(UQ(ind_quo) <= thresh_value)) %>% spread(outcome, n)
    d1$thresh_value <- thresh_value
    return(d1)
}

final_test <- thresh_measure(df, 'pass_rate', 0.8)

# alternative II
thresh_measure <- function(data, indicator, thresh_value)
{
    ind_quo <- enquo(indicator)
    d1 <- data %>%
        group_by(class_number, outcome) %>%
        summarize(n=sum(UQ(ind_quo) <= thresh_value)) %>% spread(outcome, n)
    d1$thresh_value <- thresh_value
    return(d1)
}

final_test <- thresh_measure(df, pass_rate, 0.8)