Rank variable by group (dplyr)
For future readers, the rank by group variable can be achieved using base R. Per the OP's iris
data example to rank according to Sepal.Length
:
# ORDER BY SPECIES AND SEPAL.LENGTH
iris <- iris[with(iris, order(Species, Sepal.Length)), ]
# RUN A ROW COUNT FOR RANK BY SPECIES GROUP
iris$rank <- sapply(1:nrow(iris),
function(i) sum(iris[1:i, c('Species')]==iris$Species[i]))
# FILTER DATA FRAME BY TOP 3
iris <- iris[iris$rank <= 3,]
The following produces the desired result as was specified.
library(dplyr)
by_species <- iris %>% arrange(Species, Sepal.Length) %>%
group_by(Species) %>%
mutate(rank = rank(Sepal.Length, ties.method = "first"))
by_species %>% filter(rank <= 3)
##Source: local data frame [9 x 6]
##Groups: Species [3]
##
## Sepal.Length Sepal.Width Petal.Length Petal.Width Species rank
## (dbl) (dbl) (dbl) (dbl) (fctr) (int)
##1 4.3 3.0 1.1 0.1 setosa 1
##2 4.4 2.9 1.4 0.2 setosa 2
##3 4.4 3.0 1.3 0.2 setosa 3
##4 4.9 2.4 3.3 1.0 versicolor 1
##5 5.0 2.0 3.5 1.0 versicolor 2
##6 5.0 2.3 3.3 1.0 versicolor 3
##7 4.9 2.5 4.5 1.7 virginica 1
##8 5.6 2.8 4.9 2.0 virginica 2
##9 5.7 2.5 5.0 2.0 virginica 3
by_species %>% slice(1:3)
##Source: local data frame [9 x 6]
##Groups: Species [3]
##
## Sepal.Length Sepal.Width Petal.Length Petal.Width Species rank
## (dbl) (dbl) (dbl) (dbl) (fctr) (int)
##1 4.3 3.0 1.1 0.1 setosa 1
##2 4.4 2.9 1.4 0.2 setosa 2
##3 4.4 3.0 1.3 0.2 setosa 3
##4 4.9 2.4 3.3 1.0 versicolor 1
##5 5.0 2.0 3.5 1.0 versicolor 2
##6 5.0 2.3 3.3 1.0 versicolor 3
##7 4.9 2.5 4.5 1.7 virginica 1
##8 5.6 2.8 4.9 2.0 virginica 2
##9 5.7 2.5 5.0 2.0 virginica 3