CODEWITHSUNDEEP
julia
Nick
Nick

Reputation: 9061

What's Julia's equivalent named vector in R?

In R, we could give names to each items in a vector:

> x = c( 2.718 , 3.14 , 1.414 , 47405 )            # define the vector
> names(x) = c( "e" , "pi" , "sqrt2" , "zipcode" ) # name the components
> x[c(2,4)]                   # which indices to include
pi zipcode
3.14 47405.00
> x[c(-1,-3)]                 # which indices to exclude
pi zipcode
3.14 47405.00
> x[c(FALSE,TRUE,FALSE,TRUE)] # for each position, include it?
pi zipcode
3.14 47405.00

In Julia, my first thought is using a Dict:

julia> x = [2.718, 3.14, 1.414, 47405]
4-element Array{Float64,1}:
     2.718
     3.14 
     1.414
 47405.0  

julia> namelist = ["e", "pi", "sqrt2", "zipcode"]
4-element Array{ASCIIString,1}:
 "e"      
 "pi"     
 "sqrt2"  
 "zipcode"

julia> xdict=Dict(zip(namelist, x))
Dict{ASCIIString,Float64} with 4 entries:
  "e"       => 2.718
  "zipcode" => 47405.0
  "pi"      => 3.14
  "sqrt2"   => 1.414

julia> xdict["pi"]
3.14

However, Dict lost the order of the original array, meaning I cannot access items like in R:

julia> xdict[[2,4]]
ERROR: KeyError: [2,4] not found in getindex at dict.jl:718

Is there something like named array in Julia? If not, what's the Julia-way to handle this kind of problem?

Upvotes: 5

Views: 2378

Answers (2)

Closed Limelike Curves
Closed Limelike Curves

Reputation: 190

Julia has the named tuple, which is a similar data structure that you can create like this:

(e=2.718, pi=3.14159, sqrt2 = 1.44, zipcode=47405)

(Do note that unlike named lists, named tuples are immutable. This leads to better performance, but means named tuples can be a bit of a pain to work with unless you use BangBang.jl, a package that lets you work with immutable objects as if they were mutable.)

Upvotes: 3

Dan Getz
Dan Getz

Reputation: 18227

Trying to replicate the R code using NamedArrays:

julia> using NamedArrays

julia> xx = NamedArray([2.718, 3.14, 1.414, 47405],
                       (["e", "pi", "sqrt2", "zipcode"],))
4-element NamedArrays.NamedArray...
e       2.718  
pi      3.14   
sqrt2   1.414  
zipcode 47405.0

julia> xx[[2,4]]
2-element NamedArrays.NamedArray...
pi      3.14   
zipcode 47405.0

julia> xx[setdiff(1:end,[1,3])]
2-element NamedArrays.NamedArray...
pi      3.14   
zipcode 47405.0

julia> xx[[1:end...][[false,true,false,true]]]
2-element NamedArrays.NamedArray...
pi      3.14   
zipcode 47405.0

The indexing techniques are less than optimal. Improvements are welcome in the comments. It is also possible to enhance NamedArrays easily for better indexing (which would be harder in R).

Upvotes: 6

Related Questions