BEAM Pushers Scripts

beam_by_country.exs

Mix.install([:req, :explorer])

defmodule BeamByCountry do
  require Explorer.DataFrame, as: DF

  def run do
    # Fetch languages data
    df =
      Req.get!("https://github.com/github/innovationgraph/raw/refs/heads/main/data/languages.csv").body
      |> Explorer.DataFrame.load_csv!()

    # Calculate BEAM pushers per country and quarter
    beam_by_country =
      df
      |> DF.filter(language in ["Elixir", "Erlang", "Gleam"])
      |> DF.group_by([:iso2_code, :year, :quarter])
      |> DF.summarise(beam_pushers: Explorer.Series.sum(num_pushers))

    # Calculate total BEAM pushers per quarter (exclude EU to avoid double-counting)
    beam_totals =
      beam_by_country
      |> DF.filter(iso2_code != "EU")
      |> DF.group_by([:year, :quarter])
      |> DF.summarise(total_beam_pushers: Explorer.Series.sum(beam_pushers))

    # Join and calculate ratio
    result =
      beam_by_country
      |> DF.join(beam_totals, on: [:year, :quarter], how: :left)
      |> DF.mutate(country_ratio: beam_pushers / total_beam_pushers)
      |> DF.mutate(period: cast(year, :string) <> "-Q" <> cast(quarter, :string))
      |> DF.select([:iso2_code, :period, :beam_pushers, :country_ratio])
      |> DF.sort_by([asc: iso2_code, asc: period])

    IO.puts("BEAM Developers by Country and Quarter")
    IO.puts("=======================================")
    IO.inspect(result)

    # Write to CSV
    output_path = "beam_by_country.csv"
    DF.to_csv!(result, output_path, delimiter: "\t")
    IO.puts("\nCSV written to: #{output_path}")
  end
end

BeamByCountry.run()

beam_ratio.exs

Mix.install([:req, :explorer])

defmodule BeamRatio do
  require Explorer.DataFrame, as: DF

  def run do
    # Fetch languages data
    df =
      Req.get!("https://github.com/github/innovationgraph/raw/refs/heads/main/data/languages.csv").body
      |> Explorer.DataFrame.load_csv!()

    # Calculate total pushers per quarter (exclude EU to avoid double-counting)
    totals =
      df
      |> DF.filter(iso2_code != "EU")
      |> DF.group_by([:year, :quarter])
      |> DF.summarise(total_pushers: Explorer.Series.sum(num_pushers))

    # Calculate BEAM pushers per quarter (exclude EU to avoid double-counting)
    beam =
      df
      |> DF.filter(language in ["Elixir", "Erlang", "Gleam"] and iso2_code != "EU")
      |> DF.group_by([:year, :quarter])
      |> DF.summarise(beam_pushers: Explorer.Series.sum(num_pushers))

    # Join and calculate ratio
    result =
      totals
      |> DF.join(beam, on: [:year, :quarter], how: :left)
      |> DF.mutate(beam_pushers: Explorer.Series.fill_missing(beam_pushers, 0))
      |> DF.mutate(beam_ratio: beam_pushers / total_pushers)
      |> DF.mutate(period: cast(year, :string) <> "-Q" <> cast(quarter, :string))
      |> DF.select([:period, :total_pushers, :beam_pushers, :beam_ratio])
      |> DF.sort_by([asc: period])

    IO.puts("BEAM Language Ratio by Location and Quarter")
    IO.puts("============================================")
    IO.inspect(result)

    # Write to CSV
    output_path = "beam_ratio_by_location.csv"
    DF.to_csv!(result, output_path, delimiter: "\t")
    IO.puts("\nCSV written to: #{output_path}")
  end
end

BeamRatio.run()

https://docs.google.com/spreadsheets/d/16E5HhKYeYfX6iSw3MTpM4rc2MoBZxqHu0YcHoOihgWk/edit?usp=sharing