CODEWITHSUNDEEP
arraysjulia
Femto Trader
Femto Trader

Reputation: 2014

Convert a 1-dimensional Array of objects (compound type) to a 2-dimensional Array

I have defined a compound type like

type Position
  Date::DateTime
  x::AbstractFloat
  y::AbstractFloat
  z::AbstractFloat
end

and a array 1-dimensional Array of object (compound type) is defined like this

arr = [Position(DateTime(2016,1,1,10,00,00),0,0,0),
   Position(DateTime(2016,1,1,10,00,01),2,0,0),
   Position(DateTime(2016,1,1,10,00,02),2,1,0),
   Position(DateTime(2016,1,1,10,00,03),2,2,0),
   ]

I can easily get a_Dates (array of dates) using

a_Dates = map(rec->rec.Date, arr)

but I don't know how I can get a 2-dimensional Array of positions (x, y, z)

Ideally I would like a solution without for loop. For this I know I can do (for example)

Nrows, Ncols = length(arr), length(fieldnames(Position)) - 1
data = zeros(Nrows, Ncols)

for i in 1:Nrows
    for j in 1:Ncols
        data[i, j] = getfield(arr[i], j + 1)
    end
end

Upvotes: 2

Views: 103

Answers (2)

Martin
Martin

Reputation: 11

Is [getfield(p, f) for p in arr, f in (:x, :y, :z)] what you are looking for?

(In julia 0.4, this will give an Array{Any} though, but in this case you could just prefix Float64 to get the correct type.)

Upvotes: 1

isebarn
isebarn

Reputation: 3940

Array

new_array = map(rec->[rec.x, rec.y, rec.z], arr)

additionally

new_array = vcat(new_array)
new_array = reshape(new_array,convert(Int64, length(new_array)/3),3)

or 

new_array = reshape(new_array,length(arr),3)

Upvotes: 1

Related Questions