For loop, RData, filter, and wsample in Julia

Struct

Constructors are functions that create new objects. In Julia, constructors can also be used as functions when applied to an argument tuple. This can be done using struct syntax. struct is a kind of type in Julia that is specified as a name and a set of fields. It looks similar to JSON representation without specific values assigned.

struct Foo 
    bar
    baz 
end 

# Note that when Foo is served as a constructor function,
# each element of the argument tuple is used as a value for the ordered fields.
foo = Foo("happy","bear")
foo.bar

julia> "happy"

foo.baz

julia> "bear"

# Order matters
foo = Foo("bear","happy?")
foo.bar

julia> "bear"

foo.baz 

julia> "happy?"

For loop

Creating a for loop in Julia is straightforward.The following codes will iterate over i = 1,2,3,…, 10 and print the number.

for i in 1:10
    println(i)
end 

RData

Performing simulations to find convergence using R could take a very long time. To speed things up, we want to use Julia for simulations. To do so, we load R object using RData package in Julia and use it as inputs for simulation. Follow these steps:

using RData 
import CodeBZip2 

# objs will be Dictionary
objs=load('data/pr_a2.rda')
print(objs)

julia> 
Dict{String, Any} with 1 entry:
  "pr_a2" => 9×7 DataFrame…

# extract the object to store it as a dataframe
pr_a2=objs["pr_a2"]

julia>  
9×7 DataFrame

Filter in DataFrame

To filter rows in a data frame, use filter(source => f::Function, df). The following example demonstrates filtering rows in the pr_a2 data frame using var1 and var2 to satisfy two conditions: (1) var1==s0, (2) var2==g0. Note that each row in var1 and var2 is denoted as s and g, respectively. The code would look like this:

# for each element in the column var1 and var2, 
# where each element for respective column is represented by s and g,
# find rows in pr_a2 that makes s equals s0 and g equals g0
# then return column var3
filter([:var1, :var2]=>
    (s,g)->s==s0 && g==g0, pr_a2)[:var3]

Sample

Suppose I want to generate 10 random values from a set [“a”, “b”], where “a” has a probabiility of 0.3 and “b” has a probability of 0.7. This means that the probability of drawing “a” is 0.3, and the probability of drawing “b” is 0.7. I can use wsample to perform the task.

wsample(["a", "b"], [0.3, 0.7], 10)

julia> 
10-element Vector{String}:
 "b"
 "b"
 "b"
 "b"
 "a"
 "a"
 "b"
 "b"
 "a"
 "b"

Reference

struct: https://docs.julialang.org/en/v1/manual/constructors/

RData: https://github.com/JuliaData/RData.jl

filter: https://juliadatascience.io/filter_subset

wsample: https://juliastats.org/StatsBase.jl/stable/sampling/