How do you remove un-named columns with a vector?

Question

I have a messy input file with lots of completely blank columns, and I'm trying to remove them like this (all the empty columns are un-named so R assigns V1, V2, etc.):

df1[,-grep("V\\d+", colnames(df1))]

However, the above line just returns the vector of all the empty columns (-1, -2, -3, -7, -10...), and doesn't actually remove each column the way df1[, -c(1, 2, 3, 7, 10) would.

Do I need to pass the vector differently?

Sample data (sanitized), was stored as data.table converted for dput():

structure(list(V1 = c(NA, NA, NA, NA, NA), V2 = c(NA, NA, NA, 
NA, NA), `Employee Name` = c("", "Bob", "", "Bob", "Bob"), V4 = c(NA, 
NA, NA, NA, NA), V5 = c(NA, NA, NA, NA, NA), `Question 1` = c("", 
"--", "", "Yes", ""), V7 = c(NA, NA, NA, NA, NA), V8 = c(NA, 
NA, NA, NA, NA), `Question 2` = c("", "No", "", "Yes", ""), V10 = c(NA, 
NA, NA, NA, NA), V11 = c(NA, NA, NA, NA, NA), `Question 3` = c("", 
"--", "", "Yes", ""), V13 = c(NA, NA, NA, NA, NA), V14 = c(NA, 
NA, NA, NA, NA), `Question 4` = c("", "--", "", "Yes", ""), V16 = c(NA, 
NA, NA, NA, NA), V17 = c(NA, NA, NA, NA, NA), V18 = c(NA, NA, 
NA, NA, NA), V19 = c(NA, NA, NA, NA, NA), V20 = c(NA, NA, NA, 
NA, NA), `Question 5` = c("", "--", "", "Yes", ""), V22 = c(NA, 
NA, NA, NA, NA), V23 = c(NA, NA, NA, NA, NA), V24 = c(NA, NA, 
NA, NA, NA), V25 = c(NA, NA, NA, NA, NA), `Question 6` = c("", 
"--", "", "Yes", ""), V27 = c(NA, NA, NA, NA, NA), V28 = c(NA, 
NA, NA, NA, NA), V29 = c(NA, NA, NA, NA, NA), V30 = c(NA, NA, 
NA, NA, NA), V31 = c(NA, NA, NA, NA, NA)), .Names = c("V1", "V2", 
"Employee Name", "V4", "V5", "Question 1", "V7", "V8", "Question 2", 
"V10", "V11", "Question 3", "V13", "V14", "Question 4", "V16", 
"V17", "V18", "V19", "V20", "Question 5", "V22", "V23", "V24", 
"V25", "Question 6", "V27", "V28", "V29", "V30", "V31"), row.names = c(NA, 
5L), class = "data.frame")

`df1[, -c(grep("V\\d+", colnames(df1)))]` or `df1[, -c(grep("V\\d+", colnames(df1), value = TRUE))]`? — Jaap, Jul 10 '17 at 16:10
@Jaap that throws an `invalid argument to unary operator` error — Mako212, Jul 10 '17 at 16:12
@Jaap and to clarify post your edit, `value = TRUE` throws the above error, adding the `-c` wrapper just returns the vector of column indices — Mako212, Jul 10 '17 at 16:14
Your code should work, maybe adding a [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example/5963610) would help clarify your problem. — Jaap, Jul 10 '17 at 16:18
add `with = FALSE`: `df1[,-grep("V\\d+", colnames(df1)), with = FALSE]` — Jaap, Jul 10 '17 at 16:20
You can make `class = c("data.table", "data.frame")` to have the `dput` give a data table. — Rich Scriven, Jul 10 '17 at 16:47
To be clear, the example works fine with the base R data.frame sample data. If you're using data.table, you need to **a.** show it in your example, **b.** tag it, and **c.** to solve your problem use its syntax, which differs for `[`, as Jaap alluded to above. — alistaire, Jul 10 '17 at 17:49

score 1 · Answer 1 · answered Jul 10 '17 at 16:32

The comments point out how to do the regex using grep to select the columns, but since you're using data.table you can also delete the V## columns by reference.

dat[, grep("V\\d+", colnames(dat)) := NULL]

   Employee Name Question 1 Question 2 Question 3 Question 4 Question 5 Question 6
1:                                                                                
2:           Bob         --         No         --         --         --         --
3:                                                                                
4:           Bob        Yes        Yes        Yes        Yes        Yes        Yes
5:           Bob

How do you remove un-named columns with a vector?

1 Answers1