entropic 0.1.0 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: d35ae5edab8f2a86c92301ba0a35c1c90bfd3d26
-  data.tar.gz: 3463138f664205b707808c9ae8783120258c897f
+  metadata.gz: 951873f3bb3edd267bb873b352dde384658d00ba
+  data.tar.gz: d403582aa90292e13a973be1d1652e4401266c00
 SHA512:
-  metadata.gz: 862964de635c81b16806aa8f9acb1a5e347ac399208d688ed4b5b702ddac216ad29a0f5d339c3e12d66a43eaad05c370ffdfa3bba96740a3f1f2c8b7e2aa7f4d
-  data.tar.gz: 2847b9f7d5ae73f36be387594085fd211203e9c365ee25b2cf9655b7ccd1565a9fa2b7d64961ebbb23cccebb9f889607a3b9756309dd0c69efd70f6aa848e055
+  metadata.gz: 88498c5b050afbb5e01fd6f0906d420c8b73a0e78ff1b73af2827c65838a42ad2a636363a5f455ebbed0371366cbf102b3419dfa2e4d9015ac6d53ceee970d39
+  data.tar.gz: 230990158cf17071445c4728ac8ff0317152782fc59ac30265b00bd567849b7679dad0c0ef37752576196a254ec2e5161a138be977223f2bf315f58791504584

data/README.md

@@ -8,7 +8,7 @@ require 'entropic'
 => true
 >> m.predict("entropy")
 => {:log_prob_total=>-37.181802347513745, :log_prob_average=>-6.1969670579189575, :size=>6}
->> m.predict("yportne")
+>> m.entropy("yportne")
 => {:log_prob_total=>-34.25705444264748, :log_prob_average=>-5.70950907377458, :size=>6}
 ```
 
@@ -21,8 +21,8 @@ You can also train a model, using strings one per line.
 => true
 >> File.open('/tmp/training.txt') {|f| n.train(f)}; true
 => true
->> n.predict('love')
-=> {:log_prob_total=>-15.396216763909154, :log_prob_average=>-5.132072254636385, :size=>3}
+>> n.entropy('love')
+=> 5.132072254636385
 ```
 
 You can also train a model, using strings and a count of the number of times it appers, tab separated.
@@ -32,8 +32,8 @@ You can also train a model, using strings and a count of the number of times it
 => true
 >> File.open('/tmp/training_with_counts.txt') {|f| o.train_with_multiplier(f)}; true
 => true
->> o.predict('love')
-=> {:log_prob_total=>-15.396216763909154, :log_prob_average=>-5.132072254636385, :size=>3}
+>> o.entropy('love')
+=> 5.132072254636385
 ```
 
 You can also dump a model, to be read later.

data/lib/entropic.rb

@@ -1,4 +1,4 @@
-require "entropic/version"
+require 'entropic/version'
 
 # Public: classes and methods useful for estimating entropy on strings.
 #
@@ -9,7 +9,6 @@ require "entropic/version"
 #   # => { log_prob_total: -101.1, log_prob_average: -20.02, size: 5 }
 #
 module Entropic
-
   # Public: create a sliding window of ngrams from a string
   #
   # string: The String to slide over
@@ -17,10 +16,10 @@ module Entropic
   #
   # Examples
   #
-  #  sliding('01234', 2)  
+  #  sliding('01234', 2)
   #  # => ['01', '12', '23', '34']
   #
-  def Entropic.sliding(string, n)
+  def self.sliding(string, n)
     (0..string.length - n).map { |i| (string[i, n]).to_s }
   end
 
@@ -36,7 +35,7 @@ module Entropic
     # Public: update a counter with a string, and a multiplier
     #
     # Examples
-    # 
+    #
     # counter = NGramCounter.new(2)
     # counter.update_with_multiplier('01234', 1)
     #
@@ -53,7 +52,7 @@ module Entropic
     # Public: update a counter with a string, with a multiplier of 1
     #
     # Examples
-    # 
+    #
     # counter = NGramCounter.new(2)
     # counter.update('01234')
     #
@@ -62,11 +61,11 @@ module Entropic
     def update(string)
       update_with_multiplier(string, 1)
     end
- 
+
     # Public: get count for string, with default
     #
     # Examples
-    # 
+    #
     # counter = NGramCounter.new(2)
     # counter.update('01234')
     # counter.count('01', 0)
@@ -85,18 +84,17 @@ module Entropic
   # Public; A model for entropy
   class Model
     VERSION = '1.0.0'.freeze
-    attr_accessor :size, :map
+    attr_accessor :size, :counter
 
     def initialize(size)
       @size = size
-      @map = {}
-      (1..size).each { |key| @map[key] = NGramCounter.new(key) }
+      @counter = NGramCounter.new(size)
     end
 
     # Public: update a model with a string, and a multiplier
     #
     # Examples
-    # 
+    #
     # model = Model.new(2)
     # model.update_with_multiplier('01234', 1)
     #
@@ -104,15 +102,13 @@ module Entropic
     # multiplier: The Integer describing how much weight (will often be 1)
     #
     def update_with_multiplier(string, multiplier)
-      @map.each do |_, counter|
-        counter.update_with_multiplier(string, multiplier)
-      end
+      @counter.update_with_multiplier(string, multiplier)
     end
 
     # Public: update a model with a string, with mulitplier or 1
     #
     # Examples
-    # 
+    #
     # model = Model.new(2)
     # model.update('01234')
     #
@@ -127,48 +123,37 @@ module Entropic
     # or log_prob of a 1-gram appearing once if no suffix found
     #
     # Examples
-    # 
+    #
     # model = Model.new(2)
     # model.update('01234')
     # model.log_prob('01')
-    # 
+    #
     # string: The String to query
     #
     def log_prob(key)
-      last_total = 1
-      if @map.all? { |_, m| m.counts.empty? } || !key || key == ''
+      if @counter.total == 0 || !key || key == ''
         return Math.log(0, 2.0) # -Infinity
       end
 
-      (1..key.size).each do |i|
-        k = key[-i..-1]
-        counter = @map.fetch(k.size, nil)
-        next unless counter
-        count = counter.counts.fetch(k, nil)
-        return Math.log(count, 2.0) - Math.log(counter.total, 2.0) if count
-        last_total = counter.total
-      end
-      # found it nowhere. Return '1 count' from last total
-      Math.log(1.0, 2.0) - Math.log(last_total, 2.0)
+      count = @counter.count(key, 0.5)
+      Math.log(count, 2.0) - Math.log(counter.total, 2.0)
     end
 
     # Public: dump model to some io object
-    # 
+    #
     # io: the IOWriter to write to
     #
     def dump(io)
-      @map.each do |k, m|
-        m.counts.each do |ngram, count|
-          io.write("#{k}\t#{ngram}\t#{count}\n")
-        end
+      @counter.counts.each do |ngram, count|
+        io.write("#{@size}\t#{ngram}\t#{count}\n")
       end
     end
 
     # Public: predict the log_prob sum and average over a string
     #         which will be split into ngrams
-    # 
+    #
     # string: The String to query
-    # 
+    #
     # returns: a dictionary of
     #  - log_prob_total
     #  - log_prob_average
@@ -180,35 +165,42 @@ module Entropic
       { log_prob_total: log_prob_total, log_prob_average: log_prob_average, size: ngrams.size }
     end
 
+    # Public: predict the entropy over a string
+    #         which will be split into ngrams
+    #
+    # string: The String to query
+    # 
+    # returns: entropy
+    def entropy(string)
+      -predict(string)[:log_prob_average]
+    end
+
     # Public: create a Model from reading from an IO object
     #
     # io: the IOReader
     #
     # returns: Model with stats filled in, and size of largest ngram
     def self.read(io)
-      model = Model.new(0)
-      max_size = 0
+      model = nil
       io.each_line do |string|
         ngram_size, ngram, count = string.strip.split(/\t/)
         ngram_size = ngram_size.to_i
         count = count.to_f
-        model.map[ngram_size] = NGramCounter.new(ngram_size) unless model.map.include?(ngram_size)
-        counter = model.map[ngram_size]
+        model = Model.new(ngram_size) unless model
+        counter = model.counter
         counter.total += count
         counter.counts[ngram] = count
-        max_size = ngram_size if ngram_size > max_size
       end
-      model.size = max_size
       model
     end
 
     # Public: Train a model on a bunch of data, line by line
     #
     # io: the IOReader
-    # 
+    #
     def train(io)
       io.each_line do |string|
-        update(string)
+        update(string.strip)
       end
     end
 
@@ -217,7 +209,7 @@ module Entropic
     #         each data line should be <string><tab><multiplier>
     #
     # io: the IOReader
-    # 
+    #
     def train_with_multiplier(io)
       io.each_line do |string|
         text, count = string.strip.split(/\t/)
@@ -227,4 +219,3 @@ module Entropic
     end
   end
 end
-

data/lib/entropic/version.rb

@@ -1,3 +1,3 @@
 module Entropic
-  VERSION = "0.1.0"
+  VERSION = "0.2.1"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: entropic
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.1
 platform: ruby
 authors:
 - Will Fitzgerald