say I have the iris dataset.
data(iris)
There are 150 rows in the dataset.
How can I group_by the first 50 rows and then group_by rows 51:100, and then finally group_by 101:150?
Some python code that I am following is using (nothing to do with the iris data):
data.reset_index().assign(groupId = lambda row: row.index // 1000)
A data.table approach:
library(data.table)
setDT(iris)[, .( mean( Sepal.Length ) ), by = .( rleid( 0:(nrow( iris ) - 1) %/% 50) )][]
rleid() is used to create groups to summarise by (in this case, the mean of Sepal.Length (by group of 50 rows) is calculated into column V1.
rleid V1
1: 1 5.006
2: 2 5.936
3: 3 6.588
On option would be gl (or another is rep)
library(dplyr)
iris %>%
group_by(grp = as.integer(gl(n(), 50, n()))) %>%
summarise_if(is.numeric, mean)
# A tibble: 3 x 5
# grp Sepal.Length Sepal.Width Petal.Length Petal.Width
# <int> <dbl> <dbl> <dbl> <dbl>
#1 1 5.01 3.43 1.46 0.246
#2 2 5.94 2.77 4.26 1.33
#3 3 6.59 2.97 5.55 2.03
Or another option is %/%
(seq_len(nrow(iris))-1) %/% 50 + 1
