class TestOkapi(object):
"""
Test for okapi connector
"""
df = None
random_okapi = None
popularity = None
bpr = None
def setUp(self):
"""
Setup the test package
"""
self.df = pd.read_csv(
resource_filename(testfm.__name__, "data/movielenshead.dat"),
sep="::",
header=None,
names=["user", "item", "rating", "date", "title"],
)
self.random_okapi = RandomOkapi()
self.popularity = PopularityOkapi()
self.bpr = BPROkapi()
@unittest.skipIf(not ON_REMOTE_NETWORK, "Not in igraph-01 network")
def test_get_result(self):
"""
Test if result exist behavior and get result.
"""
result_file = self.random_okapi.get_result_location(self.df)
if self.random_okapi.result_exist_for(result_file):
user, item = self.random_okapi.result
assert isinstance(user, pd.DataFrame), "First element in tuple is not a pandas DataFrame"
assert isinstance(item, pd.DataFrame), "Second element in tuple is not a pandas DataFrame"
else:
try:
result = self.random_okapi.result
except OkapiNoResultError:
pass
else:
assert False, "RandomOkapi.result is returning %s when RandomOkapi.result_exist_for is false" % result
@unittest.skipIf(not ON_REMOTE_NETWORK, "Not in igraph-01 network")
def test_map(self):
"""
Test the mapping
"""
self.random_okapi.map_data(self.df)
users = enumerate(set(self.df["user"]))
items = enumerate(set(self.df["item"]))
for user_id, user in users:
assert self.random_okapi.data_map["user_to_id"][user] == user_id, "Mapping in user to id is not correct"
assert self.random_okapi.data_map["id_to_user"][user_id] == user, "Mapping in id user is not correct"
for item_id, item in items:
assert self.random_okapi.data_map["item_to_id"][item] == item_id, "Mapping in item to id is not correct"
assert self.random_okapi.data_map["id_to_item"][item_id] == item, "Mapping in id item is not correct"
@unittest.skipIf(not ON_REMOTE_NETWORK, "Not in igraph-01 network")
def test_repeating_spliced_data(self):
"""
Test repeating data on the same model by splitting a dataFrame
"""
splitter = RandomSplitter()
for i in range(10):
training, testing = splitter.split(self.df, 0.20)
self.popularity.fit(training)
for _, row in testing.iterrows():
score = self.popularity.getScore(row["user"], row["item"])
assert isinstance(score, float), "The result of the score is not a float"
@unittest.skipIf(not ON_REMOTE_NETWORK, "Not in igraph-01 network")
def test_popularity(self):
"""
Test the popularity okapi algorithm
"""
splitter = RandomSplitter()
training, testing = splitter.split(self.df, 0.20)
pop = Popularity()
pop.fit(training)
self.popularity.fit(training)
for _, row in testing.iterrows():
assert row["user"] in training["user"]
python_score = pop.getScore(row["user"], row["item"])
okapi_score = self.popularity.getScore(row["user"], row["item"])
assert okapi_score == python_score, (
"Okapi popularity(%f) don't give the same score as his python implementation(%f)"
% (okapi_score, python_score)
)
@unittest.skipIf(not ON_REMOTE_NETWORK, "Not in igraph-01 network")
def test_bpr(self):
"""
Test the bpr okapi algorithm
"""
splitter = RandomSplitter()
training, testing = splitter.split(self.df, 0.20)
bpr = BPR()
bpr.fit(training)
self.bpr.fit(training)
for _, row in testing.iterrows():
okapi_score = self.bpr.getScore(row["user"], row["item"])
python_score = bpr.getScore(row["user"], row["item"])
assert (
okapi_score == python_score
), "Okapi bpr(%f) don't give the same score as his python implementation(%f)" % (okapi_score, python_score)