similarities

require 'rubygems'
require 'ruby-debug'

def sim_pearson(data, compare1, compare2)
  puts
  puts
  puts "sim_pearson"
  puts "==========="
  # Vemos si comparten items, sino saldremos
  shared_items = []
  data[compare1].each_key do |item|
    shared_items << item if data[compare2].has_key?(item)
  end
  n = shared_items.size
  return 0 unless n>0
  
  sum1 = sum2 = sum1Sq = sum2Sq = pSum = 0.0
  
  # puts "shared_items: #{shared_items.inspect}"
  
  shared_items.each do |item|
    data1_item = data[compare1][item] || 0.0
    data2_item = data[compare2][item] || 0.0
    puts "item: #{item} (d1item: #{data1_item} - d2item: #{data2_item})"
    sum1   += data1_item
    sum2   += data2_item
    sum1Sq += data1_item ** 2
    sum2Sq += data2_item ** 2
    pSum   += data2_item * data1_item
    # puts "  sum1: #{sum1} - sum2: #{sum2} - sum1Sq: #{sum1Sq} - sum2Sq: #{sum2Sq} - pSum: #{pSum}"
  end

  num = pSum - ((sum1*sum2)/n)
  den = Math.sqrt((sum1Sq - (sum1**2 )/n) * (sum2Sq - (sum2**2)/n))
  puts ">> num: #{num} - den: #{den} - #{(den == 0 ? 0 : num/den)}"
  return (den == 0 ? 0 : num/den)
end

def sim_blatson(data, compare1, compare2)
  puts
  puts
  puts "sim_blatson"
  puts "==========="
  # Vemos si comparten items, sino saldremos
  shared_items = []
  data[compare1].each_key do |item|
    next if !data[compare2].has_key?(item) || (data[compare1][item] == 3 && data[compare2][item] == 3)
    shared_items << item
  end
  n = shared_items.size
  return 0 unless n>0
  
  sum1 = sum2 = sum1Sq = sum2Sq = pSum = 0.0
  
  # puts "shared_items: #{shared_items.inspect}"
  
  shared_items.each do |item|
    data1_item = data[compare1][item] || 0.0
    data2_item = data[compare2][item] || 0.0
    puts "item: #{item} (d1item: #{data1_item} - d2item: #{data2_item})"
    sum1   += data1_item
    sum2   += data2_item
    sum1Sq += data1_item ** 2
    sum2Sq += data2_item ** 2
    pSum   += data2_item * data1_item
    puts "  sum1: #{sum1} - sum2: #{sum2} - sum1Sq: #{sum1Sq} - sum2Sq: #{sum2Sq} - pSum: #{pSum}"
  end

  num = pSum - ((sum1*sum2)/n)
  den = Math.sqrt((sum1Sq - (sum1**2 )/n) * (sum2Sq - (sum2**2)/n))
  puts ">> num: #{num} - den: #{den} - #{(den == 0 ? 0 : num/den)}"
  return (den == 0 ? 0 : num/den)
end

data = {
  'u1' => { 'a' => 5,   'b' => 3, 'c' => 1 },
  'u5' => { 'a' => nil, 'b' => 3, 'c' => 1 },
  'u2' => { 'a' => 1,   'b' => 3, 'c' => 1 },
  'u3' => { 'a' => 2,   'b' => 3, 'c' => 1 },
  'u4' => { 'a' => 3,   'b' => 3, 'c' => 1 },
  'u6' => { 'a' => 4,   'b' => 3, 'c' => 1 }
}

sim_pearson(data, 'u1', 'u5')
sim_pearson(data, 'u1', 'u2')
sim_pearson(data, 'u1', 'u3')
sim_pearson(data, 'u1', 'u4')
sim_pearson(data, 'u1', 'u6')
sim_blatson(data, 'u1', 'u5')
sim_blatson(data, 'u1', 'u2')
sim_blatson(data, 'u1', 'u3')
sim_blatson(data, 'u1', 'u4')
sim_blatson(data, 'u1', 'u6')

Go Blatson Go!