Reorganised tests for different models

test/simple_bayes/model/bernoulli_test.exs

@@ -0,0 +1,99 @@
+defmodule SimpleBayes.BernoulliTest do
+  use ExUnit.Case, async: true
+
+  describe "Bernoulli" do
+    # http://nlp.stanford.edu/IR-book/html/htmledition/naive-bayes-text-classification-1.html
+    test "China yes or no" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:yes, "Chinese Beijing Chinese")
+               |> SimpleBayes.train(:yes, "Chinese Chinese Shanghai")
+               |> SimpleBayes.train(:yes, "Chinese Macao")
+               |> SimpleBayes.train(:no, "Tokyo Japan Chinese")
+               |> SimpleBayes.classify("Chinese Chinese Chinese Tokyo Japan")
+
+      assert result[:yes] == 324/62500
+      assert result[:no] == 64/2916
+    end
+
+    test "binary word counting" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "red red green fruit")
+               |> SimpleBayes.train(:banana, "yellow green fruit")
+               |> SimpleBayes.train(:orange, "orange yellow fruit")
+               |> SimpleBayes.classify("red yellow fruit")
+
+      assert result[:apple] == result[:banana]
+      assert result[:apple] == result[:orange]
+    end
+
+    test "stop words" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "it is so red")
+               |> SimpleBayes.train(:banana, "it is a bit yellow")
+               |> SimpleBayes.classify_one("it is so much yellow")
+
+      assert result == :banana
+    end
+
+    test "only stop words" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "is so much")
+               |> SimpleBayes.train(:banana, "it so much")
+               |> SimpleBayes.classify("it is so much yellow")
+
+      assert result[:apple] == result[:banana]
+    end
+
+    test "smoothing - should be ignored" do
+      result = SimpleBayes.init(model: :bernoulli, smoothing: 1)
+               |> SimpleBayes.train("apple", "red")
+               |> SimpleBayes.train("banana", "yellow")
+               |> SimpleBayes.train("orange", "orange")
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result == ["apple", "orange", "banana"]
+    end
+
+    test "ordering" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "red green orange")
+               |> SimpleBayes.train(:banana, "red green")
+               |> SimpleBayes.train(:orange, "red orange")
+               |> SimpleBayes.classify("red green orange")
+               |> Keyword.keys()
+
+      assert result == [:apple, :orange, :banana]
+    end
+
+    test "keywords weighting - should be ignored" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "red", weight: 100)
+               |> SimpleBayes.train(:banana, "red", weight: 0.01)
+               |> SimpleBayes.train(:orange, "red", weight: 10)
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result != [:apple, :orange, :banana]
+    end
+
+    test "IDF (Inverse Document Frequency) - should be ignored" do
+      result = SimpleBayes.init(model: :bernoulli)
+               |> SimpleBayes.train(:apple, "red red fruit")
+               |> SimpleBayes.train(:banana, "yellow yellow fruit")
+               |> SimpleBayes.train(:orange, "orange yellow fruit")
+               |> SimpleBayes.classify("red yellow fruit")
+
+      assert result[:apple] == result[:banana]
+    end
+
+    test "stemming" do
+      result = SimpleBayes.init(model: :bernoulli, stem: true)
+               |> SimpleBayes.train(:apple, "buying apple")
+               |> SimpleBayes.train(:banana, "buy banana")
+               |> SimpleBayes.classify("buy apple")
+
+      assert result[:apple] > result[:banana]
+    end
+  end
+end

test/simple_bayes/model/binarized_multinomial_test.exs

@@ -0,0 +1,86 @@
+defmodule SimpleBayes.BinarizedMultinomialTest do
+  use ExUnit.Case, async: true
+
+  describe "Binarized multinomial" do
+    test "binary word counting" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "red red green fruit")
+               |> SimpleBayes.train(:banana, "yellow green fruit")
+               |> SimpleBayes.train(:orange, "orange yellow fruit")
+               |> SimpleBayes.classify("red yellow fruit")
+
+      assert result[:apple] == result[:banana]
+      assert result[:apple] == result[:orange]
+    end
+
+    test "stop words" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "it is so red")
+               |> SimpleBayes.train(:banana, "it is a bit yellow")
+               |> SimpleBayes.classify_one("it is so much yellow")
+
+      assert result == :banana
+    end
+
+    test "only stop words" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "is so much")
+               |> SimpleBayes.train(:banana, "it so much")
+               |> SimpleBayes.classify("it is so much yellow")
+
+      assert result[:apple] == result[:banana]
+    end
+
+    test "smoothing - should be ignored" do
+      result = SimpleBayes.init(model: :binarized_multinomial, smoothing: 1)
+               |> SimpleBayes.train("apple", "red")
+               |> SimpleBayes.train("banana", "yellow")
+               |> SimpleBayes.train("orange", "orange")
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result == ["apple", "orange", "banana"]
+    end
+
+    test "ordering" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "red green orange")
+               |> SimpleBayes.train(:banana, "red green")
+               |> SimpleBayes.train(:orange, "red orange")
+               |> SimpleBayes.classify("red green orange")
+               |> Keyword.keys()
+
+      assert result == [:apple, :orange, :banana]
+    end
+
+    test "keywords weighting - should be ignored" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "red", weight: 100)
+               |> SimpleBayes.train(:banana, "red", weight: 0.01)
+               |> SimpleBayes.train(:orange, "red", weight: 10)
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result != [:apple, :orange, :banana]
+    end
+
+    test "IDF (Inverse Document Frequency) - should be ignored" do
+      result = SimpleBayes.init(model: :binarized_multinomial)
+               |> SimpleBayes.train(:apple, "red red fruit")
+               |> SimpleBayes.train(:banana, "yellow yellow fruit")
+               |> SimpleBayes.train(:orange, "orange yellow fruit")
+               |> SimpleBayes.classify("red yellow fruit")
+
+      assert result[:apple] == result[:banana]
+    end
+
+    test "stemming" do
+      result = SimpleBayes.init(model: :binarized_multinomial, stem: true)
+               |> SimpleBayes.train(:apple, "buying apple")
+               |> SimpleBayes.train(:banana, "buy banana")
+               |> SimpleBayes.classify("buy apple")
+
+      assert result[:apple] > result[:banana]
+    end
+  end
+end

test/simple_bayes/model/multinomial_test.exs

@@ -0,0 +1,64 @@
+defmodule SimpleBayes.MultinomialTest do
+  use ExUnit.Case, async: true
+
+  describe "multinomial" do
+    test "stop words" do
+      result = SimpleBayes.init
+               |> SimpleBayes.train(:apple, "it is so red")
+               |> SimpleBayes.train(:banana, "it is a bit yellow")
+               |> SimpleBayes.classify_one("it is so much yellow")
+
+      assert result == :banana
+    end
+
+    test "only stop words" do
+      result = SimpleBayes.init
+               |> SimpleBayes.train(:apple, "is so much")
+               |> SimpleBayes.train(:banana, "it so much")
+               |> SimpleBayes.classify("it is so much yellow")
+
+      assert result[:apple] == result[:banana]
+    end
+
+    test "smoothing" do
+      result = SimpleBayes.init(smoothing: 1)
+               |> SimpleBayes.train("apple", "red")
+               |> SimpleBayes.train("banana", "yellow")
+               |> SimpleBayes.train("orange", "orange")
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result == ["orange", "banana", "apple"]
+    end
+
+    test "ordering" do
+      result = SimpleBayes.init
+               |> SimpleBayes.train(:apple, "red", weight: 100)
+               |> SimpleBayes.train(:banana, "red", weight: 0.01)
+               |> SimpleBayes.train(:orange, "red", weight: 10)
+               |> SimpleBayes.classify("red")
+               |> Keyword.keys()
+
+      assert result == [:apple, :orange, :banana]
+    end
+
+    test "IDF (Inverse Document Frequency)" do
+      result = SimpleBayes.init
+               |> SimpleBayes.train(:apple, "red red fruit")
+               |> SimpleBayes.train(:banana, "yellow yellow fruit")
+               |> SimpleBayes.train(:orange, "orange yellow fruit")
+               |> SimpleBayes.classify("red yellow fruit")
+
+      assert result[:apple] > result[:banana]
+    end
+
+    test "stemming" do
+      result = SimpleBayes.init(stem: true)
+               |> SimpleBayes.train(:apple, "buying apple")
+               |> SimpleBayes.train(:banana, "buy banana")
+               |> SimpleBayes.classify("buy apple")
+
+      assert result[:apple] > result[:banana]
+    end
+  end
+end