Drop columns that match a pattern, with an exception-CodePudding

I have example data as follows:

dat <- mtcars
dat$add <- dat[,1]
dat$last <- dat[,1]
names(dat) <- c("fraction_low", "fraction_medium", "fraction_high", "fraction_low_2000", "fraction_medium_2000", "fraction_high_2000","fraction_low_2001", "fraction_medium_2001", "fraction_high_2001","fraction_low_comb", "fraction_medium_comb", "fraction_high_comb", "last")

I would like to remove all columns/variable names that have fraction and low, medium, or high in their name, except the columns/variable names that also have comb in the name.

I found a lot of good answers for removing columns with a certain pattern here. For example:

library(dplyr)
dat %>% select( -( contains("fraction") & ( (contains("low") | contains("medium") | contains("high") ) )))

But how should I implement and exception to such a pattern?

Desired output:

desired_output <- dat[,10:13]

                    fraction_low_comb fraction_medium_comb fraction_high_comb last
Mazda RX4                           4                    4               21.0 21.0
Mazda RX4 Wag                       4                    4               21.0 21.0
Datsun 710                          4                    1               22.8 22.8
Hornet 4 Drive                      3                    1               21.4 21.4

CodePudding user response：

We may use

library(dplyr)
 dat %>%
   select(-matches(c("fraction", "low", "high", "medium")), contains("comb"))

Or may use

dat %>% 
   select(-matches("fraction_(low|medium|high)_(?!comb)|fraction_(low|medium|high)$",
     perl = TRUE))