Consider a (m x n) matrix of only 0s and 1s, with m potentially large.
julia> rand([0, 1], 5, 3)
5×3 Array{Int64,2}:
0 1 1
0 0 0
0 1 1
1 0 0
1 0 1
Is there an efficient way to count the number of occurrences and track the indices for each unique row?
For example, the first row above occurs twice, at indices 1 and 3. I am trying to build a sort of contingency table.
Thanks
This is one of the approaches that is based only on functionalities provided in Julia Base:
julia> x = rand([0, 1], 20, 3)
20×3 Matrix{Int64}:
1 0 1
1 1 1
0 0 0
1 0 0
0 0 1
1 0 1
0 0 0
1 0 1
0 0 0
0 0 1
1 1 0
0 1 1
0 1 1
1 0 0
0 0 0
0 0 0
0 0 1
0 1 0
1 1 0
1 0 0
julia> d = Dict()
Dict{Any, Any}()
julia> for (i, r) in enumerate(eachrow(x))
push!(get!(d, r, Int[]), i)
end
julia> d
Dict{Any, Any} with 8 entries:
[1, 1, 1] => [2]
[0, 0, 0] => [3, 7, 9, 15, 16]
[0, 0, 1] => [5, 10, 17]
[1, 1, 0] => [11, 19]
[1, 0, 0] => [4, 14, 20]
[0, 1, 1] => [12, 13]
[1, 0, 1] => [1, 6, 8]
[0, 1, 0] => [18]
and now using the SplitApplyCombine.jl package:
julia> using SplitApplyCombine
julia> group(i -> view(x, i, :), axes(x, 1))
8-element Dictionaries.Dictionary{Any, Vector{Int64}}
[1, 0, 1] │ [1, 6, 8]
[1, 1, 1] │ [2]
[0, 0, 0] │ [3, 7, 9, 15, 16]
[1, 0, 0] │ [4, 14, 20]
[0, 0, 1] │ [5, 10, 17]
[1, 1, 0] │ [11, 19]
[0, 1, 1] │ [12, 13]
[0, 1, 0] │ [18]