How do I select variables in an R dataframe whose names contain a particular string?

Question

Two examples would be very helpful for me.

How would I select: 1) variables whose names start with b or B (i.e. case-insensitive) or 2) variables whose names contain a 3

df <- data.frame(a1 = factor(c("Hi", "Med", "Hi", "Low"), 
  levels = c("Low", "Med", "Hi"), ordered = TRUE),
  a2 = c("A", "D", "A", "C"), a3 = c(8, 3, 9, 9),
  b1 = c(1, 1, 1, 2), b2 = c( 5, 4, 3,2), b3 = c(3, 4, 3, 4),
  B1 = c(3, 6, 4, 4))

Answer 1

If you just want the variable names:

grep("^[Bb]", names(df), value=TRUE)

grep("3", names(df), value=TRUE)

If you are wanting to select those columns, then either

df[,grep("^[Bb]", names(df), value=TRUE)]
df[,grep("^[Bb]", names(df))]

The first uses selecting by name, the second uses selecting by a set of column numbers.

Answer 2

While I like the answer above, I wanted to give a "tidyverse" solution as well. If you are doing a lot of pipes and trying to do several things at once, as I often do, you may like this answer. Also, I find this code more "humanly" readable.

The function tidyselect::vars_select will select variables from a character vector in the first argument, which should contain the names of the corresponding data frame, based on a select helper function like starts_with or matches

library(dplyr)
library(tidyselect)


df <- data.frame(a1 = factor(c("Hi", "Med", "Hi", "Low"), 
                         levels = c("Low", "Med", "Hi"), ordered = TRUE),
             a2 = c("A", "D", "A", "C"), a3 = c(8, 3, 9, 9),
             b1 = c(1, 1, 1, 2), b2 = c( 5, 4, 3,2), b3 = c(3, 4, 3, 4),
             B1 = c(3, 6, 4, 4))

# will select the names starting with a "b" or a "B"
tidyselect::vars_select(names(df), starts_with('b', ignore.case = TRUE)) 

# use select in conjunction with the previous code
df %>%
  select(vars_select(names(df), starts_with('b', ignore.case = TRUE)))

# Alternatively
tidyselect::vars_select(names(df), matches('^[Bb]'))

Note that the default for ignore.case is TRUE, but I put it here to show explicitly, and in case future readers are curious how to adjust the code. The include and exclude arguments are also very useful. For example, you could use vars_select(names(df), matches('^[Bb]'), include = 'a1') if you wanted everything that starts with a "B" or a "b", and you wanted to include "a1" as well.