How would I automate dropping a column in R based on summary data for that column?

Question

I have a dataset that is being used to create an automated dashboard. Essentially it's looking at the relationship between certain conditions and the cost of care on a month by month basis for a health care institution. What I want to be able to do is in pseudocode:

dataset %>% select(-c("columns where the average value is lower than X"))

No amount of googling seems to be getting me close.

Answer 1

We can use select_if

library(dplyr)
val <- 10
dataset %>%
    select_if(~ is.numeric(.) && mean(.) < val)

---

Or using base R

dataset[, names(which(colMeans(dataset[sapply(dataset, class) == 
            "numeric"]) < val)), drop = FALSE]
#   col3
#1    3
#2    4
#3    7

data

dataset <- data.frame(col1 = c('A', 'B', 'C'), col2 = c(10, 8, 15),
     col3 = c(3, 4, 7), stringsAsFactors = FALSE)