Select nth element in data frame by factor

Question

I've got a dataframe with a text column name and factor city. It is ordered alphabetically firstly by city and then name. Now I need to get a data frame which contains only nth element in each city, keeping this ordering. How can it be done in a pretty way without loops?

I have:

name    city
John    Atlanta
Josh    Atlanta
Matt    Atlanta
Bob     Boston
Kate    Boston
Lily    Boston
Matt    Boston

I want a function, which returns n'th element by city, i.e., if it is 3rd, then:

name    city
Matt    Atlanta
Lily    Boston

It should return NULL for name if it is out of range for the selected city, i.e., for 4th:

name    city
NULL    Atlanta
Matt    Boston

Using only base R please?

Answer 1

In base R using by:

Set up some test data, including an additional out of range value:

test <- read.table(text="name    city
John    Atlanta
Josh    Atlanta
Matt    Atlanta
Bob     Boston
Kate    Boston
Lily    Boston
Matt    Boston
Bob     Seattle
Kate    Seattle",header=TRUE)

Get the 3rd item in each city:

do.call(rbind,by(test,test$city,function(x) x[3,]))

Result:

        name    city
Atlanta Matt Atlanta
Boston  Lily  Boston
Seattle <NA>    <NA>

To get exactly what you want, here is a little function:

nthrow <- function(dset,splitvar,n) {
    result <- do.call(rbind,by(dset,dset[splitvar],function(x) x[n,]))
    result[,splitvar][is.na(result[,splitvar])] <- row.names(result)[is.na(result[,splitvar])]
    row.names(result) <- NULL
    return(result)
}

Call it like:

nthrow(test,"city",3)

Result:

  name    city
1 Matt Atlanta
2 Lily  Boston
3 <NA> Seattle

Answer 2

A data.table solution

library(data.table)
DT <- data.table(test)

# return all columns from the subset data.table
n <- 4
DT[,.SD[n,] ,by = city]
##      city name
## 1: Atlanta   NA
## 2:  Boston Matt
## 3: Seattle   NA

# if you just want the nth element of `name` 
# (excluding other columns that might be there)
# any of the following would work

DT[,.SD[n,] ,by = city, .SDcols = 'name']


DT[, .SD[n, list(name)], by = city]


DT[, list(name = name[n]), by = city]