y_true[: len(evaluation_set_read)] = user_model.READ
y_true[len(evaluation_set_read) :] = user_model.UNREAD
# Set y_pred
y_pred = np.empty(shape=(y_true.shape[0]))
predicted_interesting_headlines = list()
actual_interesting_headlines = list()
# predict with other subset and record measures
for i, article_id in enumerate(evaluation_set_read):
article = Article.objects(id=article_id).first()
if article is None:
continue
# Predict and record result
result = user_model.rank(doc=article)
y_pred[i] = result
# Redcord headline
if result == user_model.READ:
actual_interesting_headlines.append(article.headline)
actual_interesting_headlines.append(article.headline)
for i, article_id in enumerate(evaluation_set_unread, start=len(evaluation_set_read)):
article = Article.objects(id=article_id).first()
if article is None:
continue
# Predict and record result
result = user_model.rank(doc=article)
class UserModelSVMTest(unittest.TestCase):
def setUp(self):
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.DEBUG)
config = load_config(file_path="/vagrant/config.yaml", logger=logger, exit_with_error=True)
#Connect to test database
connect("nyan_test", port=27017)
fill_database()
#connect(config['database']['db-name'],
# username= config['database']['user'],
# password= config['database']['passwd'],
# port = config['database']['port'])
self.user_id = User.objects(email=u'*****@*****.**').first().id
#feature_extractor = EsaFeatureExtractor(prefix = config['prefix'])
feature_extractor = TfidfFeatureExtractor(prefix=config['prefix'])
self.trainer = UserModelSVM(self.user_id, extractor=feature_extractor)
def tearDown(self):
clear_database()
def test_mean_std_deviation(self):
X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [3, 0], [3, 2]])
self.trainer._calculate_mean_and_std_deviation(X)
self.assertAlmostEqual(self.trainer.theta_[0], 0.16666, 4)
self.assertAlmostEqual(self.trainer.theta_[1], -0.16666, 4)
self.assertAlmostEqual(self.trainer.sigma_[0], 2.33927, 4)
self.assertAlmostEqual(self.trainer.sigma_[1], 1.3437, 4)
def test_normalize(self):
X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [3, 0], [3, 2]])
self.trainer.theta_ = np.array([1, -2], dtype=np.float32)
self.trainer.sigma_ = np.array([2, 1], dtype=np.float32)
X = self.trainer._normalize(X)
self.assertEqual(X[1, 0], -1.5)
def test_normalize_no_theta(self):
X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [3, 0], [3, 2]])
self.assertRaises(AttributeError, lambda: self.trainer._normalize(X))
#dummy set theta_ but not sigma_
self.trainer.theta_ = np.array([1, -2], dtype=np.float32)
self.assertRaises(AttributeError, lambda: self.trainer._normalize(X))
def test_save_load_theta_sigma(self):
X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [3, 0], [3, 2]])
self.trainer._calculate_mean_and_std_deviation(X)
self.trainer.clf = "dummy"
tmp_theta = self.trainer.theta_
tmp_sigma = self.trainer.sigma_
self.trainer.save()
self.trainer.load()
#Check normalization parameters
self.assertEqual(tmp_theta.all(),
self.trainer.theta_.all())
self.assertEqual(tmp_sigma.all(),
self.trainer.sigma_.all())
@unittest.skip("No ranking implemented yet")
def test_rank(self):
self.trainer.train()
unread_doc = Article.objects(headline=u"Sony = Bad").first()
read_doc = Article.objects(headline=u"Apple").first()
rank_unread_doc = self.trainer.rank(unread_doc)
rank_read_doc = self.trainer.rank(read_doc)
self.assertEqual(rank_unread_doc, UserModelBayes.UNREAD)
self.assertEqual(rank_read_doc, UserModelBayes.READ)
