Bradley terry values can be calculated, mapped to score

Dhruv Kapadia
1 parent c01226e1
Showing 4 changed files with 174 additions and 0 deletions Show diff stats
app/models/choice.rb
app/models/question.rb
lib/tasks/import_bt_test_data.rb
lib/tasks/makecsvfromhash.rb
@@ -82,6 +82,26 @@ class Choice &lt; ActiveRecord::Base
     self.item.creator_id != self.question.creator_id
   end
  
+  def compute_bt_score(btprobs = nil)
+      if btprobs.nil?
+	      btprobs = self.question.bradley_terry_probs
+      end
+
+      p_i = btprobs[self.id]
+
+      total = 0
+      btprobs.each do |id, p_j|
+	      if id == self.id
+		      next
+	      end
+
+	      total += (p_i / (p_i + p_j))
+      end
+
+      total / (btprobs.size-1)
+
+  end
+
  
   protected
  
@@ -93,6 +93,73 @@ class Question &lt; ActiveRecord::Base
 	   weighted.each_with_index {|item, i| weighted[i] = item/sum}
    end
   end
+
+ def bradley_terry_probs
+   probs = []
+   prev_probs = []
+
+   # What ordering key we use is unimportant, just need a consistent way to link index of prob to id
+   the_choices = self.choices.sort{|x,y| x.id<=>y.id}
+
+   # This hash is keyed by pairs of choices - 'LC.id, RC.id'
+   the_prompts = prompts_hash_by_choice_ids
+
+   # Initial probabilities chosen at random
+   the_choices.size.times do 
+	   probs << rand
+	   prev_probs << rand
+   end
+
+   t=0
+   probs_size = probs.size
+   
+   # probably want to add a fuzz here to account for floating rounding
+   until probs == prev_probs do
+      s = t % probs_size
+      prev_probs = probs.dup
+      choice = the_choices[s]
+
+      numerator = choice.wins.to_f
+
+      denominator = 0.0
+      the_choices.each_with_index do |c, index|
+	      if(index == s)
+		      next
+	      end
+
+	      wins_and_losses = the_prompts["#{choice.id}, #{c.id}"].votes.size + the_prompts["#{c.id}, #{choice.id}"].votes.size
+	      denominator+= (wins_and_losses).to_f / (prev_probs[s] + prev_probs[index])
+      end
+      probs[s] = numerator / denominator 
+      normalize!(probs)
+      t+=1
+   end
+ 
+   probs_hash = {}
+   probs.each_with_index do |item, index| 
+	   probs_hash[the_choices[index].id] = item
+   end
+   probs_hash
+ end
+
+ 
+ def all_bt_scores
+	 btprobs = bradley_terry_probs
+	 btprobs.each do |key, value|
+		 c = Choice.find(key)
+		 puts "#{c.id}: #{c.votes.size} #{c.compute_bt_score(btprobs)}"
+	 end
+
+ end
+
+ def prompts_hash_by_choice_ids
+   the_prompts = {}
+   self.prompts.each do |p|
+      the_prompts["#{p.left_choice_id}, #{p.right_choice_id}"] = p
+   end
+   the_prompts
+ end
+
  
    def distinct_array_of_choice_ids(rank = 2, only_active = true)
      @choice_ids = choice_ids
@@ -0,0 +1,62 @@
+require 'faster_csv'
+
+BASEDIR = "/home/dhruv/CITP/bt_test/"
+
+q = Question.new(:name => "test for bt", :creator_id => 1, :site_id => 1)
+q.save()
+
+choice_offset = Choice.last.id
+
+inserts = []
+
+timestring = Time.now.to_s(:db) #isn't rails awesome?
+totalchoices=0
+FasterCSV.foreach(BASEDIR + "choices_7000.txt", {:headers => :first_row, :return_headers => false}) do |choice|
+# for each choice, create an insert with unique id
+   id = choice[0].to_i + choice_offset
+   wins = choice[1].to_i
+   inserts.push("(#{q.id}, #{id}, #{wins}, '#{timestring}', '#{timestring}')")
+   totalchoices+=1
+end
+
+sql = "INSERT INTO `choices` (`question_id`, `item_id`, `votes_count`, `created_at`, `updated_at`) VALUES #{inserts.join(', ')}"
+
+ActiveRecord::Base.connection.execute(sql)
+
+inserts = []
+prompt_offset = Prompt.last.id
+totalprompts = 0
+FasterCSV.foreach(BASEDIR + "prompts_7000.txt", {:headers => :first_row, :return_headers => false}) do |prompt|
+   id = prompt[0].to_i + prompt_offset
+   left_choice_id = prompt[1].to_i + choice_offset
+   right_choice_id = prompt[2].to_i + choice_offset
+   votes_count =  prompt[3].to_i
+
+   inserts.push("(NULL, #{q.id}, NULL, #{left_choice_id}, '#{timestring}', '#{timestring}', NULL, #{votes_count}, #{right_choice_id}, NULL, NULL)")
+   totalprompts +=1
+end
+
+sql = "INSERT INTO `prompts` (`algorithm_id`, `question_id`, `voter_id`, `left_choice_id`, `created_at`, `updated_at`, `tracking`, `votes_count`, `right_choice_id`, `active`, `randomkey`) VALUES #{inserts.join(', ')}"
+
+ActiveRecord::Base.connection.execute(sql)
+
+inserts = []
+vote_offset = Vote.last.id
+totalvotes=0
+FasterCSV.foreach(BASEDIR + "votes_7000.txt", {:headers => :first_row, :return_headers => false}) do |vote|
+   id = vote[0].to_i + vote_offset
+   prompt_id = vote[1].to_i + prompt_offset 
+   choice_id = vote[2].to_i + choice_offset
+   loser_choice_id = vote[3].to_i 
+
+   inserts.push("(#{prompt_id}, #{q.id}, #{choice_id}, #{loser_choice_id}, '#{timestring}', '#{timestring}')")
+   totalvotes +=1
+end
+
+sql = "INSERT INTO `votes` (`prompt_id`, `question_id`, `choice_id`, `loser_choice_id`, `created_at`, `updated_at`) VALUES #{inserts.join(', ')}"
+
+ActiveRecord::Base.connection.execute(sql)
+
+sql = "UPDATE questions SET votes_count=#{totalvotes}, prompts_count=#{totalprompts}, choices_count=#{totalchoices} WHERE id=#{q.id}"
+
+ActiveRecord::Base.connection.execute(sql)
@@ -0,0 +1,25 @@
+require 'fastercsv'
+
+q = Question.find(109)
+
+the_prompts = q.prompts_hash_by_choice_ids
+
+#hash_of_choice_ids_from_left_to_right_to_votes
+the_hash = {}
+the_prompts.each do |key, p|
+	left_id, right_id = key.split(", ")
+        if not the_hash.has_key?(left_id)
+	   the_hash[left_id] = {}
+	   the_hash[left_id][left_id] = 0
+	end
+
+        the_hash[left_id][right_id] = p.votes.size
+end
+
+the_hash.sort.each do |xval, row|
+	rowarray = []
+	row.sort.each do |yval, cell|
+		rowarray << cell
+	end
+	puts rowarray.join(", ")
+end
...	...	@@ -82,6 +82,26 @@ class Choice < ActiveRecord::Base
82	82	self.item.creator_id != self.question.creator_id
83	83	end
84	84
	85	+ def compute_bt_score(btprobs = nil)
	86	+ if btprobs.nil?
	87	+ btprobs = self.question.bradley_terry_probs
	88	+ end
	89	+
	90	+ p_i = btprobs[self.id]
	91	+
	92	+ total = 0
	93	+ btprobs.each do \|id, p_j\|
	94	+ if id == self.id
	95	+ next
	96	+ end
	97	+
	98	+ total += (p_i / (p_i + p_j))
	99	+ end
	100	+
	101	+ total / (btprobs.size-1)
	102	+
	103	+ end
	104	+
85	105
86	106	protected
87	107
...	...
...	...	@@ -93,6 +93,73 @@ class Question < ActiveRecord::Base
93	93	weighted.each_with_index {\|item, i\| weighted[i] = item/sum}
94	94	end
95	95	end
	96	+
	97	+ def bradley_terry_probs
	98	+ probs = []
	99	+ prev_probs = []
	100	+
	101	+ # What ordering key we use is unimportant, just need a consistent way to link index of prob to id
	102	+ the_choices = self.choices.sort{\|x,y\| x.id<=>y.id}
	103	+
	104	+ # This hash is keyed by pairs of choices - 'LC.id, RC.id'
	105	+ the_prompts = prompts_hash_by_choice_ids
	106	+
	107	+ # Initial probabilities chosen at random
	108	+ the_choices.size.times do
	109	+ probs << rand
	110	+ prev_probs << rand
	111	+ end
	112	+
	113	+ t=0
	114	+ probs_size = probs.size
	115	+
	116	+ # probably want to add a fuzz here to account for floating rounding
	117	+ until probs == prev_probs do
	118	+ s = t % probs_size
	119	+ prev_probs = probs.dup
	120	+ choice = the_choices[s]
	121	+
	122	+ numerator = choice.wins.to_f
	123	+
	124	+ denominator = 0.0
	125	+ the_choices.each_with_index do \|c, index\|
	126	+ if(index == s)
	127	+ next
	128	+ end
	129	+
	130	+ wins_and_losses = the_prompts["#{choice.id}, #{c.id}"].votes.size + the_prompts["#{c.id}, #{choice.id}"].votes.size
	131	+ denominator+= (wins_and_losses).to_f / (prev_probs[s] + prev_probs[index])
	132	+ end
	133	+ probs[s] = numerator / denominator
	134	+ normalize!(probs)
	135	+ t+=1
	136	+ end
	137	+
	138	+ probs_hash = {}
	139	+ probs.each_with_index do \|item, index\|
	140	+ probs_hash[the_choices[index].id] = item
	141	+ end
	142	+ probs_hash
	143	+ end
	144	+
	145	+
	146	+ def all_bt_scores
	147	+ btprobs = bradley_terry_probs
	148	+ btprobs.each do \|key, value\|
	149	+ c = Choice.find(key)
	150	+ puts "#{c.id}: #{c.votes.size} #{c.compute_bt_score(btprobs)}"
	151	+ end
	152	+
	153	+ end
	154	+
	155	+ def prompts_hash_by_choice_ids
	156	+ the_prompts = {}
	157	+ self.prompts.each do \|p\|
	158	+ the_prompts["#{p.left_choice_id}, #{p.right_choice_id}"] = p
	159	+ end
	160	+ the_prompts
	161	+ end
	162	+
96	163
97	164	def distinct_array_of_choice_ids(rank = 2, only_active = true)
98	165	@choice_ids = choice_ids
...	...
...	...	@@ -0,0 +1,62 @@
	1	+require 'faster_csv'
	2	+
	3	+BASEDIR = "/home/dhruv/CITP/bt_test/"
	4	+
	5	+q = Question.new(:name => "test for bt", :creator_id => 1, :site_id => 1)
	6	+q.save()
	7	+
	8	+choice_offset = Choice.last.id
	9	+
	10	+inserts = []
	11	+
	12	+timestring = Time.now.to_s(:db) #isn't rails awesome?
	13	+totalchoices=0
	14	+FasterCSV.foreach(BASEDIR + "choices_7000.txt", {:headers => :first_row, :return_headers => false}) do \|choice\|
	15	+# for each choice, create an insert with unique id
	16	+ id = choice[0].to_i + choice_offset
	17	+ wins = choice[1].to_i
	18	+ inserts.push("(#{q.id}, #{id}, #{wins}, '#{timestring}', '#{timestring}')")
	19	+ totalchoices+=1
	20	+end
	21	+
	22	+sql = "INSERT INTO `choices` (`question_id`, `item_id`, `votes_count`, `created_at`, `updated_at`) VALUES #{inserts.join(', ')}"
	23	+
	24	+ActiveRecord::Base.connection.execute(sql)
	25	+
	26	+inserts = []
	27	+prompt_offset = Prompt.last.id
	28	+totalprompts = 0
	29	+FasterCSV.foreach(BASEDIR + "prompts_7000.txt", {:headers => :first_row, :return_headers => false}) do \|prompt\|
	30	+ id = prompt[0].to_i + prompt_offset
	31	+ left_choice_id = prompt[1].to_i + choice_offset
	32	+ right_choice_id = prompt[2].to_i + choice_offset
	33	+ votes_count = prompt[3].to_i
	34	+
	35	+ inserts.push("(NULL, #{q.id}, NULL, #{left_choice_id}, '#{timestring}', '#{timestring}', NULL, #{votes_count}, #{right_choice_id}, NULL, NULL)")
	36	+ totalprompts +=1
	37	+end
	38	+
	39	+sql = "INSERT INTO `prompts` (`algorithm_id`, `question_id`, `voter_id`, `left_choice_id`, `created_at`, `updated_at`, `tracking`, `votes_count`, `right_choice_id`, `active`, `randomkey`) VALUES #{inserts.join(', ')}"
	40	+
	41	+ActiveRecord::Base.connection.execute(sql)
	42	+
	43	+inserts = []
	44	+vote_offset = Vote.last.id
	45	+totalvotes=0
	46	+FasterCSV.foreach(BASEDIR + "votes_7000.txt", {:headers => :first_row, :return_headers => false}) do \|vote\|
	47	+ id = vote[0].to_i + vote_offset
	48	+ prompt_id = vote[1].to_i + prompt_offset
	49	+ choice_id = vote[2].to_i + choice_offset
	50	+ loser_choice_id = vote[3].to_i
	51	+
	52	+ inserts.push("(#{prompt_id}, #{q.id}, #{choice_id}, #{loser_choice_id}, '#{timestring}', '#{timestring}')")
	53	+ totalvotes +=1
	54	+end
	55	+
	56	+sql = "INSERT INTO `votes` (`prompt_id`, `question_id`, `choice_id`, `loser_choice_id`, `created_at`, `updated_at`) VALUES #{inserts.join(', ')}"
	57	+
	58	+ActiveRecord::Base.connection.execute(sql)
	59	+
	60	+sql = "UPDATE questions SET votes_count=#{totalvotes}, prompts_count=#{totalprompts}, choices_count=#{totalchoices} WHERE id=#{q.id}"
	61	+
	62	+ActiveRecord::Base.connection.execute(sql)
...	...
...	...	@@ -0,0 +1,25 @@
	1	+require 'fastercsv'
	2	+
	3	+q = Question.find(109)
	4	+
	5	+the_prompts = q.prompts_hash_by_choice_ids
	6	+
	7	+#hash_of_choice_ids_from_left_to_right_to_votes
	8	+the_hash = {}
	9	+the_prompts.each do \|key, p\|
	10	+ left_id, right_id = key.split(", ")
	11	+ if not the_hash.has_key?(left_id)
	12	+ the_hash[left_id] = {}
	13	+ the_hash[left_id][left_id] = 0
	14	+ end
	15	+
	16	+ the_hash[left_id][right_id] = p.votes.size
	17	+end
	18	+
	19	+the_hash.sort.each do \|xval, row\|
	20	+ rowarray = []
	21	+ row.sort.each do \|yval, cell\|
	22	+ rowarray << cell
	23	+ end
	24	+ puts rowarray.join(", ")
	25	+end
...	...