Add empty columns to a dataframe with specified names from a vector

Question

I have a dataframe, df, with a a number of columns of data already. I have a vector, namevector, full of strings. I need empty columns added to df with the names of the columns from namevector.

I am trying to add columns with this for loop, iterating over each string in namevector.

for (i in length(namevector)) {
  df[, i] <- NA
}

but am left with this error:

Error in [<-.data.frame(*tmp*, , i, value = NA) : new columns would leave holes after existing columns

Alternatively, I have thought of creating an empty dataframe with the correct names, then cbind-ing the two dataframes together but am not sure how to go about coding this.

How would I go about resolving this?

I think maybe you meant `seq_len(length(namevector))` or something similar. — joran, Aug 13 '13 at 16:33
you mean `for (i in (seq_len(length(namevector)) + ncol(df))) ...` then change the column names. — James Pringle, Aug 13 '13 at 17:03
@JamesPringle `seq_len(length(...))` is a longcut to `seq_along()` — Señor O, Aug 13 '13 at 17:06
@JamesPringle Thanks, I forgot about adding the offset since they're adding new columns. — joran, Aug 13 '13 at 17:09

score 85 · Accepted Answer · answered Aug 13 '13 at 17:10

85

The problem with your code is in the line:

for(i in length(namevector))

You need to ask yourself: what is length(namevector)? It's one number. So essentially you're saying:

for(i in 11)
df[,i] <- NA

Or more simply:

df[,11] <- NA

That's why you're getting an error. What you want is:

for(i in namevector)
    df[,i] <- NA

Or more simply:

df[,namevector] <- NA

answered Aug 13 '13 at 17:10

Señor O

17,049
2
45
47

9

Please consider moving `df[,namevector] <- NA` to the top of your answer, since that is the general solution – stevec Feb 12 '21 at 00:13

dayne · Answer 2 · 2017-09-20T11:58:57.387

31

set.seed(1)
example <- data.frame(col1 = rnorm(10, 0, 1), col2 = rnorm(10, 2, 3))
namevector <- c("col3", "col4")
example[ , namevector] <- NA

example
#          col1       col2 col3 col4
# 1  -0.6264538  6.5353435   NA   NA
# 2   0.1836433  3.1695297   NA   NA
# 3  -0.8356286  0.1362783   NA   NA
# 4   1.5952808 -4.6440997   NA   NA
# 5   0.3295078  5.3747928   NA   NA
# 6  -0.8204684  1.8651992   NA   NA
# 7   0.4874291  1.9514292   NA   NA
# 8   0.7383247  4.8315086   NA   NA
# 9   0.5757814  4.4636636   NA   NA
# 10 -0.3053884  3.7817040   NA   NA

edited Sep 20 '17 at 11:58

answered Aug 13 '13 at 16:57

dayne

7,504
6
38
56

The `set.seed(1)` doesn't seem strictly necessary. – Faheem Mitha Jan 30 '15 at 16:12
4

The `set.seed` was only added so the example data.frame will be the same if a user runs the code. Technically, the `rnorm` is not necessary at all. – dayne Jan 30 '15 at 19:22
1

Yes, so the random numbers are the same for everyone runs the code. I see. – Faheem Mitha Jan 30 '15 at 21:53

score 16 · Answer 3 · edited Oct 08 '18 at 13:11

16

The below works for me

dataframe[,"newName"] <- NA

Make sure to add "" for new name string.

edited Oct 08 '18 at 13:11

David Arenburg

91,361
17
137
196

answered Feb 11 '16 at 17:55

Shubham

434
6
12

The OP has requested to *Add empty columns to a dataframe with specified names **from a vector***. So, he wants to add *many* columns. Your answer does only add *one* column and the name is *not* taken from the variable `namevector` as requested. Therefore, please review your post or consider to delete it. Thank you. – Uwe Oct 09 '18 at 05:35
you can always run a loop for list of columns – Shubham Jan 28 '19 at 23:38
Right but then your answer is identical to the [accepted answer](https://stackoverflow.com/a/18215216/3817004) which suggests `for(i in namevector) df[,i] <- NA` – Uwe Jan 28 '19 at 23:43

score 4 · Answer 4 · answered Oct 10 '22 at 21:47

I prefer the base R solution, but here is a tidyverse solution. Below uses mutate, but the syntax is the same if you were to use tibble::add_column:

library(dplyr)

new_columns <- c("A", "B", "C")

iris %>% 
  mutate(!!!setNames(rep(NA, length(new_columns)), new_columns))

The splice operator uses the name-value pair from setNames to create the columns.

If you had name-value pairs in a list already then you can simply do:

new_columns <- list(A = NA, B = 1, C = "c")

iris %>% 
  mutate(!!!new_columns)

If the length of each list element is 1 then its value is recycled. Otherwise, it needs to have the same length as the data frame (e.g. list(A = 1:150).

Agriculturist · Answer 5 · 2023-01-05T16:45:02.617

A different way to add empty columns to a dataframe.

df <- data.frame(name0 = c(1,2,3))
namevector <- c("name1", "name2", "name3")

sapply(namevector, function(x){df[x] <<- numeric()})

The for loop is eliminated with the <<- operator.
In many of these examples, the type associated with the new empty column will be a logical; however, by specifying numeric() the type will be numeric. Other types, like double() or character(), could be specified.

score -2 · Answer 6 · answered Aug 13 '13 at 16:53

-2

Maybe

df <- do.call("cbind", list(df, rep(list(NA),length(namevector))))
colnames(df)[-1*(1:(ncol(df) - length(namevector)))] <- namevector

answered Aug 13 '13 at 16:53

James Pringle

1,079
6
15

Add empty columns to a dataframe with specified names from a vector

6 Answers6

Linked

Related