Subset Returns 0 Rows

Question

I am new to R. I tried to pull data from a data frame A using subset Data frame A looks like this:

col a       col b
1            1     
1           NA
NA          NA
1           1

I want to find out the group with col a = 1 and col b <> 1 My code:

test <- subset(A, A$a == 1 & A$b == NULL)

OR

test <- subset(A, A$a == 1 & A$b <> 1)

test returns 0 rows.

sum(is.na(A$a))

results: 5126

sum(is.na(A$b))

results: 6753

What is better ways to pull data using R?

the not-equals operator is `!=`, and to test for `NA` you use `is.na(...)` (and therefore, for not-NA use `!is.na(...)`) — SymbolixAU, Oct 02 '16 at 21:45
so for your first example, `subset(A, A$a == 1 & is.na(A$b))` — SymbolixAU, Oct 02 '16 at 21:47
@SymbolixAU Thanks! it works. any idea why my previous trials failed? — Ninjia123, Oct 02 '16 at 21:50
Firslty, `NULL` and `NA` are two different beasts (e.g. [here](http://stackoverflow.com/q/15496361/5977215) ), secondly, to test for `NULL` or `NA` use `is.na()` and `is.null()` respectively. For example, see the difference between these tests: `nullTest <- c(NULL);nullTest == NULL;is.null(nullTest)` — SymbolixAU, Oct 02 '16 at 21:53
And when you have `NA`s in your data, you can't check for them by testing if they are not-equal to a number (so your `A$b != 1` won't find the NAs, you need to explicitly test for NA) — SymbolixAU, Oct 02 '16 at 21:55

score 1 · Accepted Answer · edited May 23 '17 at 12:06

1

To answer the "What is better ways to pull data using R" part of your question: you sould avoid using subset as it can cause problems and cannot be used to assign values. This has been discussed there:

Why is `[` better than `subset`?

and there:

http://www.cookbook-r.com/Basics/Getting_a_subset_of_a_data_structure/

In my opinion it is better to use [ or learn the data.table package.

edited May 23 '17 at 12:06

Community

1
1

answered Oct 03 '16 at 13:41

GuillaumeL

985
8
11

Subset Returns 0 Rows

1 Answers1