Problems with NAs

user_895124 0
Viewed 107

I have a data set like

dat
   ejer_id person_alder koen  aar
1        1            9    1 2011
2        2            9    1 2011
3        3            7    1 2011
4        4           94    1 2011
5        5           94    2 2011
6        6           93   NA 2011
7        7           93    1 2011
8        8           91    2 2011
9        9           91    1 2011
10      10           91   NA 2011

I count the NAs:

isna <- sum(is.na(dat$koen))

which gives

> isna
   [1] 2

I don't understand why the following doesn't work:

 > length( dat$koen[dat$koen == 1] )
[1] 8

This should be 6.

length( dat$koen[dat$koen == 2] ) [1] 4

This should be 2.

I can get the following to work:

> length( which( dat$koen == 1 ) )
[1] 6
> length( which( dat$koen == 2 ) )
[1] 2

My data set is quite large, so I need to know, what I am doing, and I don't understand the difference between these two expressions.

Any help is greatly appreciated.

na r
2 Answers

Remove NA's, following code help you:

length(na.omit(dat$koen[dat$koen==1]))

Or same can be done in following way:

length(dat$koen[which(!is.na(dat$koen[dat$koen==1]))])

if this wiil not work, explain your issue properly

user_29313640

The thing you should know is that you cannot compare nothing (NA) with something (1 here)., i.e. when the algorithm does this NA==1, it does NOT return FALSE but it will return NA. That is the reason you are getting:

 > length( dat$koen[dat$koen == 1] )
[1] 8

Try this instead:

 > length(dat$koen[dat$koen==1 & !is.na(dat$koen)])
[1] 6
user_27875860
Related