Probability calculator

laplacian-smoothing.rb

require 'set'

k = 1.0

list1 = %w(a perfect world my perfect woman pretty woman)
list2 = %w(a perfect day electric storm another rainy day)
ncases1 = 3
ncases2 = 3
list1name = 'movie'
list2name = 'song'
words = %w(perfect storm)

# list1 = %w(offer is secret click secret link secret sports link)
# list2 = %w(play sports today went play sports secret sports event sports is today sports costs money)
# ncases1 = 3
# ncases2 = 5
# list1name = 'spam'
# list2name = 'ham'
# words = %w(today is secret)

x = Set.new list1
x.merge list2

n1 = list1.length.to_f
n2 = list2.length.to_f
ncases = ncases1 + ncases2

plist1 = (ncases1 + k) / (ncases + 2 * k)
plist2 = (ncases2 + k) / (ncases + 2 * k)
puts "P(#{list1name}) = #{plist1}"
puts "P(#{list2name}) = #{plist2}"

pword1 = {}
pword2 = {}
words.each do |w|
  pword1[w] = (list1.count(w).to_f + k) / (n1 + k * x.size)
  pword2[w] = (list2.count(w).to_f + k) / (n2 + k * x.size)
  puts "P(\"#{w}\"|#{list1name}) = #{pword1[w]}"
  puts "P(\"#{w}\"|#{list2name}) = #{pword2[w]}"
end

factor1 = words.inject(1.0) {|a,e| a * pword1[e]} * plist1
factor2 = words.inject(1.0) {|a,e| a * pword2[e]} * plist2
puts "P(#{list1name}|\"#{words.join(' ')}\") = #{factor1 / (factor1 + factor2)}"
puts "P(#{list2name}|\"#{words.join(' ')}\") = #{factor2 / (factor1 + factor2)}"

linear-regression.rb

module Inject
  def sum
    self.inject(0) {|a,e| a + e}
  end
end

class Array
  include Inject
end

x = [0,1,2,3,4].collect &:to_f
y = [3,6,7,8,11].collect &:to_f

xy = x.zip(y).collect {|a,b| a*b}
x2 = x.collect {|e| e*e}
y2 = y.collect {|e| e*e}
m = x.length.to_f
w1 = (m * xy.sum - x.sum * y.sum) / (m * x2.sum - x.sum * x.sum)
w0 = y.sum / m - w1 / m * x.sum

puts "w0 = #{w0}, w1 = #{w1}"

maximum-likelihood.rb

require 'set'

list1 = %w(a perfect world my perfect woman pretty woman)
list2 = %w(a perfect day electric storm another rainy day)
ncases1 = 3
ncases2 = 3
list1name = 'movie'
list2name = 'song'
words = %w(perfect storm)

# list1 = %w(offer is secret click secret link secret sports link)
# list2 = %w(play sports today went play sports secret sports event sports is today sports costs money)
# ncases1 = 3
# ncases2 = 5
# list1name = 'spam'
# list2name = 'ham'
# words = %w(today is secret)

x = Set.new list1
x.merge list2

n1 = list1.length.to_f
n2 = list2.length.to_f
ncases = (ncases1 + ncases2).to_f

plist1 = ncases1 / ncases
plist2 = ncases2 / ncases
puts "P(#{list1name}) = #{plist1}"
puts "P(#{list2name}) = #{plist2}"

pword1 = {}
pword2 = {}
words.each do |w|
  pword1[w] = list1.count(w).to_f / n1
  pword2[w] = list2.count(w).to_f / n2
  puts "P(\"#{w}\"|#{list1name}) = #{pword1[w]}"
  puts "P(\"#{w}\"|#{list2name}) = #{pword2[w]}"
end

factor1 = words.inject(1.0) {|a,e| a * pword1[e]} * plist1
factor2 = words.inject(1.0) {|a,e| a * pword2[e]} * plist2
puts "P(#{list1name}|\"#{words.join(' ')}\") = #{factor1 / (factor1 + factor2)}"
puts "P(#{list2name}|\"#{words.join(' ')}\") = #{factor2 / (factor1 + factor2)}"