def test_item_user_model(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
self.assertEqual(tfidf._get_item_models(self.df), {1: ["oh", "god"], 100: ["car", "very", "nice"],
110: ["sky", "sky", "blue", "nice"]})
self.assertEqual(tfidf._users, {10: set([100]), 11: set([100, 1]), 12: set([110])})
self.assertEqual(sorted(tfidf.tfidf.keys()), sorted([100, 1, 110]))
def test_item_user_model(self):
tfidf = TFIDFModel('desc')
tfidf.fit(self.df)
self.assertEqual(tfidf._get_item_models(self.df), {1: ['oh', 'god'],
100: ['car', 'very', 'nice'],
110: ['sky', 'sky', 'blue', 'nice']})
self.assertEqual(tfidf._users, {10: set([100]), 11: set([100, 1]), 12: set([110])})
self.assertEqual(sorted(tfidf.tfidf.keys()), sorted([100, 1, 110]))
def test_item_model(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
self.assertAlmostEqual(tfidf._sim(1, 1), 1, places=2)
self.assertAlmostEqual(tfidf._sim(100, 100), 1, places=2)
self.assertGreater(tfidf._sim(1, 1), tfidf._sim(1, 100),
"similarities do not make sense")
def test_item_user_model(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
self.assertEqual(
tfidf._get_item_models(self.df), {
1: ["oh", "god"],
100: ["car", "very", "nice"],
110: ["sky", "sky", "blue", "nice"]
})
self.assertEqual(tfidf._users, {
10: set([100]),
11: set([100, 1]),
12: set([110])
})
self.assertEqual(sorted(tfidf.tfidf.keys()), sorted([100, 1, 110]))
def test_item_model(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
self.assertAlmostEqual(tfidf._sim(1, 1), 1, places=2)
self.assertAlmostEqual(tfidf._sim(100, 100), 1, places=2)
self.assertGreater(tfidf._sim(1, 1), tfidf._sim(1, 100), "similarities do not make sense")
def test_get_score(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
tfidf.k = 1
#the closes item to 1 (in user 10 profile) is 100, so the score should be equal to the similarity
self.assertAlmostEqual(tfidf.get_score(10, 1),
tfidf._sim(100, 1),
places=2)
def test_get_score(self):
tfidf = TFIDFModel("desc")
tfidf.fit(self.df)
tfidf.k = 1
#the closes item to 1 (in user 10 profile) is 100, so the score should be equal to the similarity
self.assertAlmostEqual(tfidf.get_score(10, 1), tfidf._sim(100, 1), places=2)
