def test_NaiveBayesClassifier_with_smoothing():
model = Model(simple_tokenizer(SPAM_DATA), k=1)
verify(model.classes["spam"].prior, 0.4)
verify(model.classes["ham"].prior, 0.6)
verify(model.classes["spam"].P("today"), 0.0476)
verify(model.classes["ham"].P("today"), 0.1111)
classifier = Classifier(model)
verify(classifier.get_normalised_probabilities(["today", "is", "secret"])["spam"], 0.4858)
def test_NaiveBayesClassifier_without_smoothing():
model = Model(simple_tokenizer(SPAM_DATA), k=0)
assert model.features_num == 12
verify(model.classes["spam"].prior, 0.3750)
verify(model.classes["spam"].P("secret"), 0.3333)
verify(model.classes["ham"].P("secret"), 0.0667)
classifier = Classifier(model)
verify(classifier.get_normalised_probabilities(["sports"])["spam"], 0.1667)
verify(classifier.get_normalised_probabilities(["secret", "is", "secret"])["spam"], 0.9615)
verify(classifier.get_normalised_probabilities(["today", "is", "secret"])["spam"], 0)
def test_NaiveBayesClassifier_with_smoothing_2():
model = Model(simple_tokenizer(TITLE_DATA), k=1)
assert model.features_num == 11
verify(model.classes["movie"].prior, 0.5000)
verify(model.classes["song"].prior, 0.5000)
verify(model.classes["movie"].P("perfect"), 0.1579)
verify(model.classes["song"].P("perfect"), 0.1053)
verify(model.classes["movie"].P("storm"), 0.0526)
verify(model.classes["song"].P("storm"), 0.1053)
classifier = Classifier(model)
verify(classifier.get_normalised_probabilities(["perfect", "storm"])["movie"], 0.4286)
def test_NaiveBayesClassifier_without_smoothing_2():
model = Model(simple_tokenizer(TITLE_DATA), k=0)
classifier = Classifier(model)
verify(classifier.get_normalised_probabilities(["perfect", "storm"])["movie"], 0.0)