def test_normalize_dataset_should_return_a_dataframe(self):
users_json = b'{"users": [{"user_id": 1, "movies": [1, 2, 3, 4]}]}'
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
assert isinstance(nrecommender._normalize_dataset(), pd.DataFrame)
assert nrecommender._normalize_dataset().shape == (4, 3)
def test_get_movies(self):
expected_json = b'{"movies": [{"1": "Test"}, {"2": "Test2"}]}'
temp_file = self.get_temp_file(expected_json)
nrecommender = MoviesRecommender(temp_file.name)
assert (
nrecommender._get_movies() == json.loads(expected_json)['movies'])
def test_should_recommend_k_movies(self):
users_json = b'''
{
"users": [
{"user_id": 1, "movies": [1, 2, 3, 4]},
{"user_id": 2, "movies": [3, 2, 4]}
],
"movies": {
"1": "Toy Story (1995)",
"2": "GoldenEye (1995)",
"3": "Twelve Monkeys (1995)",
"4": "Richard III (1995)",
"5": "Seven (Se7en) (1995)",
"6": "Usual Suspects, The (1995)"
}
}
'''
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
movie_id = 2
movies = nrecommender.recommend(movie_id, 4)
""" Based on id 2 it will pick up the id that has correlation
in other users, until the k """
assert sorted(movies) == sorted([
'GoldenEye (1995)', 'Richard III (1995)', 'Twelve Monkeys (1995)',
'Toy Story (1995)'
])
def test_should_create_a_sparse_matrix(self):
users_json = b'''
{"users": [
{"user_id": 1, "movies": [1, 2, 3, 4]},
{"user_id": 2, "movies": [3, 2, 4]}
]}
'''
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
sparse_data = nrecommender._get_sparse_data()
assert isinstance(sparse_data, np.ndarray)
assert sparse_data.shape == (2, 4)
def test_should_get_correlation_item(self):
users_json = b'''
{"users": [
{"user_id": 1, "movies": [1, 2, 3, 4]},
{"user_id": 2, "movies": [3, 2, 4]}
]}
'''
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
sparse_data = nrecommender._get_sparse_data()
train, _ = nrecommender._create_train_test_data(sparse_data)
item_correlation = nrecommender._get_item_correlation(train)
assert item_correlation.shape == (4, 4)
assert not (item_correlation == 0).all()
def test_should_create_train_test_data(self):
users_json = b'''
{"users": [
{"user_id": 1, "movies": [1, 2, 3, 4]},
{"user_id": 2, "movies": [3, 2, 4]}
]}
'''
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
sparse_data = nrecommender._get_sparse_data()
train, test = nrecommender._create_train_test_data(sparse_data)
assert isinstance(train, np.ndarray)
assert train.shape == (2, 4)
assert isinstance(test, np.ndarray)
assert test.shape == (2, 4)
def test_should_return_top_k_movies(self):
users_json = b'''
{
"users": [
{"user_id": 1, "movies": [1, 2, 3, 4]},
{"user_id": 2, "movies": [3, 2, 4]}
],
"movies": {
"1": "Toy Story (1995)",
"2": "GoldenEye (1995)",
"3": "Twelve Monkeys (1995)",
"4": "Richard III (1995)",
"5": "Seven (Se7en) (1995)",
"6": "Usual Suspects, The (1995)"
}
}
'''
temp_file = self.get_temp_file(users_json)
nrecommender = MoviesRecommender(temp_file.name)
sparse_data = nrecommender._get_sparse_data()
train, _ = nrecommender._create_train_test_data(sparse_data)
item_correlation = nrecommender._get_item_correlation(train)
movie_id = 2
for i in range(1, 5):
movies = nrecommender._get_top_k_movies(item_correlation, movie_id,
i)
assert len(movies) == i
def test_get_file(self):
temp_file = self.get_temp_file(b'{"test": {"test1": 100}}')
nrecommender = MoviesRecommender(temp_file.name)
assert nrecommender