Any applications of Julia in Official Statistics?

I tried to find something without success.
There is a site with many applications in official statistics but mainly in R, Python and Java: awesome-official-statistics-software

I don’t know too much about “creating and accessing official statistics” but it it’s just a matter of getting access to some data, then you can call all Python, R and Java libraries (or call Julia from those languages), and even NodeJS libraries, such as those under the “Access to official statistics (GSBPM 7.4)” heading.

See here (a lot of Julia stats packages are available):

some info might be outdated, e.g. regarding:

The New York Fed was one of the first Julia users, if that applies.

I don’t know what the major data format is, but if this one (not just JSON, that works fine in Julia), then the package is outdated and will not install in Julia 1.0+ so you may need to use PyCall or RCall and get data that way, while it should also be simple to update this package to work (e.g. add Projects.toml file):

(@v1.6) pkg> add

I confirmed pyjstat works in Python and then the same example in Julia:

using PyCall

you just have to get the dataset over to Julia (there’s probably a better way straight into a Julia JSON, that keeps order of the keys, DIct doesn’t, or DataFrame were such applies):

julia> julia_side = py""
Dict{Any, Any} with 8 entries:
  "source"    => "Self-elaboration"
  "class"     => "dataset"
  "id"        => ["place of birth", "age group", "gender", "year", "province of residence", "concept"]
  "dimension" => Dict{Any, Any}("age group"=>Dict{Any, Any}("label"=>"age group", "category"=>Dict{Any, Any}("label"=>Dict{Any, Any}("0-4"=>"0-4", "75-79"=>"75-79", "…
  "value"     => Union{Nothing, Float64}[2.69588e6, 1.09603e6, 357648.0, 338446.0, 903759.0, 2.77293e6, 1.14129e6, 348067.0, 328697.0, 954877.0  …  7.0, 0.0, 0.0, 3.0…
  "size"      => [6, 22, 3, 2, 5, 1]
  "version"   => "2.0"
  "updated"   => "2021-03-28T18:30:26.556933"

Depending on what you mean by statistics and “official”, you can’t have a larger dataset to do statistics on than the entire Universe, i.e. “performs approximate Bayesian inference”:

Celeste: A New Model for Cataloging the Universe. This research collaboration of astrophysicists, statisticians and

1 Like