Multivariate statistics

multivariate_statistics

# frozen_string_literal: true

# rubocop:disable Lint/MissingCopEnableDirective
# rubocop:disable Layout/LineLength

require 'json'
input = [[9, 2], [4, 4], [7, 3], [9, 7], [0, 7]] # Change this as you need

def get_x(input)
  input.map { |(x, _y)| x }
end

def get_y(input)
  input.map { |(_x, y)| y }
end

def rank_array(array)
  python = <<~PYTHON
    python -c "from scipy.stats import rankdata; print(list(rankdata(#{array})))"
  PYTHON

  JSON.parse(`#{python}`).map(&:to_r).map { |r| r.denominator == 1 ? r.to_i : r }
end

def display_ranking(rank)
  o = rank.map { |r| r.is_a?(Rational) ? "#{r.numerator}/#{r.denominator}" : r }.join(',')

  "[#{o}]"
end

def mean(arr)
  (1.0 / arr.length) * arr.sum
end

def covariance(arr, x_mean = nil, y_mean = nil)
  x_mean ||= mean(get_x(arr))
  y_mean ||= mean(get_y(arr))

  (1.0 / arr.length) * arr.sum { |(x, y)| (x - x_mean) * (y - y_mean) }
end

def standard_deviation(arr)
  mean = mean(arr)

  Math.sqrt((1.0 / arr.length.to_f) * (arr.sum { |x| (x - mean)**2 }))
end

def pearson(covariance, sx, sy)
  covariance / (sx * sy)
end

# Process input data
x = get_x(input)
y = get_y(input)

x_sorted = x.sort
y_sorted = y.sort

x_rank = rank_array(x)
y_rank = rank_array(y)

# Classify correlation
input_covariance = covariance(input).round(2)

sx = standard_deviation(x)
sy = standard_deviation(y)

rank_covariance = covariance(x_rank.zip(y_rank)).round(2)

# Output evaluation
[
  'Eingabeevaluation: ',
  '-' * 30,
  "X: #{x}",
  "Y: #{y}",
  "Sortierte X: #{x_sorted}",
  "Sortierte Y: #{y_sorted}",
  "Rangliste x: #{display_ranking(x_rank)}",
  "Rangliste y: #{display_ranking(y_rank)}",
  '',
  'Auswertung: ',
  '-' * 30,
  "Kovarianz der Eingabe: #{input_covariance}",
  "Sx (Standardabweichung): #{sx}",
  "Sy (Standardabweichung): #{sy}",
  "Pearson: #{pearson(input_covariance, sx, sy).round(2)}",
  "Kovarianz der Ranglisten: #{rank_covariance}",
  "Spearman Korrelationskoeffizient: #{pearson(rank_covariance, standard_deviation(x_rank), standard_deviation(y_rank)).round(2)}"
].each(&method(:puts))

Requirements:

• Ruby (obviously)
• Python executable in PATH
• scipy installed