matthieugomez / GroupedArrays.jl

Installation

The package is registered in the General registry and so can be installed at the REPL with

] add GroupedArrays.

Introduction

GroupedArray returns an AbstractArray with integers corresponding to each group (or a missing for groups with missing).

using GroupedArrays
p = repeat(["a", "b", missing], outer = 2)
g = GroupedArray(p)
# 6-element GroupedArray{Int64, 1}:
#  1
#  2
#   missing
#  1
#  2
#   missing

Use the keyword argument coalesce = true to consider missing values as distinct

using GroupedArrays
p = repeat(["a", "b", missing], outer = 2)
g = GroupedArray(p; coalesce = true)
# 6-element GroupedArray{Int64, 1}:
#  1
#  2
#  3
#  1
#  2
#  3

GroupedArray can be used to compute groups across multiple vectors:

p1 = repeat(["a", "b"], outer = 3)
p2 = repeat(["d", "e"], inner = 3)
g = GroupedArray(p1, p2)
# 6-element GroupedArray{Int64, 1}:
#  1
#  2
#  1
#  3
#  4
#  3

Motivation

GroupedArrays is similar to PooledArray, except that the pool is simply the set of integers from 1 to n where n is the number of groups(missing is encoded as 0). This allows for faster lookup in setups where the group value is not meaningful.

See also

The algorithm to construct GroupedArrays is taken from DataFrames.jl