Clean Uruguay names & gender dataset

convert.rb

# encoding: UTF-8

require 'csv'

filename = 'nombre_nacim_por_anio_y_sexo.csv'

class Name
  attr_reader :name, :gender, :male_count, :female_count, :year

  def self.valid?(name)
    !!name
  end

  def initialize(name, gender, freq, year=1960)
    @year = year
    @name = format(name)
    @gender = translate(gender)
    @male_count = 0
    @female_count = 0
    if female?
      @female_count += freq.to_i
    elsif male?
      @male_count += freq.to_i
    end
  end

  def format(cad)
    cad = cad.strip.downcase
    cad[0] = cad[0].upcase
    cad
  end

  def translate(str)
    if str && str.strip.downcase == 'femenino'
      'female'
    elsif str && str.strip.downcase == 'masculino'
      'male'
    else
      'unknown'
    end
  end

  def female?
    gender == 'female'
  end

  def male?
    gender == 'male'
  end

  def ==(other)
    return @name == other.name && @gender == other.gender
  end

  def merge(other)
    @male_count += other.male_count
    @female_count += other.female_count
  end

  def <=>(other)
    return @name <=> other.name
  end

  def to_s
    "<Name: #{@name} | #{@gender} | #{@freq}>"
  end

  def to_row
    #Name,years.appearing,count.male,count.female,prob.gender,obs.male,est.male,upper,lower
    [@name, year_appearing, male_count, female_count, prob_gender, nil, nil, nil, nil]
  end

  def year_appearing
    2014 - year
  end

  def prob_gender
    if male_count > female_count
      'male'
    elsif female_count > male_count
      'female'
    else
      'unknown'
    end
  end
end

names = []
CSV.foreach(filename, "r:windows-1252") do |row|
  if Name.valid?(row[2])
    new_name = Name.new(row[2], row[1], row[3])
    old_name = names.detect { |name| name == new_name }
    if old_name
      old_name.merge(new_name)
    else
      names << new_name
    end
  end
end

CSV.open("output_monte.csv", "w:iso-8859-1") do |csv|
  names.sort.each { |name| csv << name.to_row }
end