def test_spam_probability(self):
training_set = [("foo bar", 1), ("bar baz", 0)]
counts = nb.count_words(training_set)
word_probs = nb.word_probabilities(counts, 1, 1)
message = "there is no foo bar like bar bar"
actual = nb.spam_probability(word_probs, message)
expected = 0.9
self.assertAlmostEqual(actual, expected)
def train(self, training_set):
# count spam and non-spam messages
num_spams = len([is_spam
for message, is_spam in training_set # pylint: disable=unused-variable
if is_spam])
num_non_spams = len(training_set) - num_spams
# run training data through our "pipeline"
word_counts = nb.count_words(training_set)
self.word_probs = nb.word_probabilities(word_counts,
num_spams,
num_non_spams,
self.k)
def test_word_probabilities(self):
training_set = [("foo bar", 1), ("bar baz", 0)]
counts = nb.count_words(training_set)
actual = nb.word_probabilities(counts, 1, 1)
expected = [("baz", 0.25, 0.75), ("foo", 0.75, 0.25), ("bar", 0.75, 0.75)]
self.assertEqual(actual, expected)
def test_count_words(self):
training_set = [("foo bar", 1)]
actual = nb.count_words(training_set)
expected = {"foo": [1, 0], "bar": [1, 0]}
self.assertEqual(actual, expected)