CODEWITHSUNDEEP
r
mtrc
mtrc

Reputation: 1345

Counting the number of entries as a means of filtering data

Sorry if the description is vague - I'm extremely new to R and finding it hard to visualise exactly what I want to do. Suppose I have some data:

dat <- read.table(text = '
A    B    C
"Mike"    1    1
"Mike"    1    17
"Mike"    1    3
"Mike"    2    4
"Mike"    3    18
"Simon"    1    2
"Simon"    1    25
"Simon"    2    12
"Simon"    2    182
"Simon"    2    6', header=TRUE)
... etc.

Suppose I want to know the number of names (A column) that have 3 entries where B = 1, and the number of names that have 3 entries where B = 2, and so on?

In the example above, "Mike" has 3 entries where B = 1, but not B = 2 or B = 3. "Simon" has 3 entries for B = 2, and so on. It's crossing entries in the data, which I've not done yet in R, and I'm not sure how best to approach it.

Upvotes: 2

Views: 86

Answers (2)

IRTFM
IRTFM

Reputation: 263411

Assuming this is in a data.frame named dat:

> tapply(dat$B, dat$A, function(x) names(table(x))[table(x)==3] )
 Mike Simon 
  "1"   "2"

Your comment suggest you wanted a tabular display. So perhaps this would also be of interest:

> xtabs( ~ A + B, dat)
       B
A       1 2 3
  Mike  3 1 1
  Simon 2 3 0

And there are methods of working with that matrix that are sometimes what is needed:

> which( xtabs( ~ A + B, dat) == 3, arr.ind=TRUE )
      row col
Mike    1   1
Simon   2   2

Upvotes: 3

Josh O&#39;Brien
Josh O&#39;Brien

Reputation: 162401

I believe this is what you're after (but realize the code's terribly dense for an R newbie, and possibly even for not-so-newbies):

tab <- table(dat[1:2])
m <- max(tab)
apply(rbind(tab, m), 2, tabulate) - c(rep(0, m-1), 1)
#      1 2 3
# [1,] 0 1 1
# [2,] 1 0 0
# [3,] 1 1 0

Values of B are along the top while frequencies (number of people having that count of B=1, B=2, and B=3) are along the side.

Upvotes: 1

Related Questions