def test_fit(self):
"""
[recommendation.models.TensorCoFi] Test size of matrix after tensorCoFi fit
"""
tf = TensorCoFi(n_users=len(self.df.user.unique()), n_items=len(self.df.item.unique()), n_factors=2)
tf.fit(self.df)
#item and user are row vectors
#self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
def load_rest():
User.load_to_cache()
# Load main models
Popularity.load_to_cache()
TensorCoFi.load_to_cache()
if "recommendation.language" in settings.INSTALLED_APPS:
from recommendation.language.models import Region
Region.load_to_cache()
if "recommendation.diversity" in settings.INSTALLED_APPS:
from recommendation.diversity.models import ItemGenre, Genre
Genre.load_to_cache()
ItemGenre.load_to_cache()
def setup_class(cls, *args, **kwargs):
"""
Put elements in db
"""
for app in ITEMS:
Item.objects.create(**app)
for u in USERS:
user = User.objects.create(external_id=u["external_id"])
for i in u["items"]:
Inventory.objects.create(user=user, item=Item.get_item_by_external_id(i))
TensorCoFi.train_from_db()
Popularity.train_from_db()
TensorCoFi.load_to_cache()
Popularity.load_to_cache()
def test_tensor_score_against_testfm(self):
"""
[recommendation.models.TensorCoFi] Test tensorcofi scores with test.fm benchmark
"""
evaluator = Evaluator()
tc = TensorCoFi(n_users=len(self.df.user.unique()), n_items=len(self.df.item.unique()), n_factors=2)
ptc = PyTensorCoFi()
training, testing = testfm.split.holdoutByRandom(self.df, 0.9)
items = training.item.unique()
tc.fit(training)
ptc.fit(training)
tc_score = evaluator.evaluate_model(tc, testing, all_items=items)[0]
ptc_score = evaluator.evaluate_model(ptc, testing, all_items=items)[0]
assert abs(tc_score-ptc_score) < .15, \
"TensorCoFi score is not close enough to testfm benchmark (%.3f != %.3f)" % (tc_score, ptc_score)
def get_model(self):
"""
Catch model
:return: The Model
"""
return TensorCoFi.get_model_from_cache()
def setup_class(cls, *args, **kwargs):
"""
Put elements in db
"""
path = resource_filename(recommendation.__name__, "/")
fill.FillTool({"items": True, "--mozilla": True, "prod": True}).load()
fill.FillTool({"users": True, "--mozilla": True, "<path>": path+"data/user"}).load()
modelcrafter.main("train", "popularity")
modelcrafter.main("train", "tensorcofi")
# Load user and items
Item.load_to_cache()
User.load_to_cache()
# Load main models
Popularity.load_to_cache()
TensorCoFi.load_to_cache()
cls.client = Client()
def test_training(self):
"""
[recommendation.models.TensorCoFi] Test train from database
"""
try:
TensorCoFi.train_from_db()
except Exception:
assert False, "Training is not working for jumping ids"
TensorCoFi.load_to_cache()
t = TensorCoFi.get_model_from_cache()
for user in User.objects.all():
if len(user.owned_items) > 2:
assert isinstance(t.get_recommendation(user), np.ndarray), "Recommendation is not a numpy array"
else:
try:
t.get_recommendation(user)
except KeyError:
pass
else:
assert False, "User with less than 3 items give a static recommendation"
def test_recommendation_with_testfm(self):
"""
[recommendation.api.GetRecommendation] Test recommendation with testfm
"""
data = np.array(zip(*map(lambda x: (x["user_id"]-1, x["item_id"]-1, 1.),
Inventory.objects.all().values("user_id", "item_id"))), dtype=np.float32)
users, items = zip(*Inventory.objects.all().values_list("user_id", "item_id"))
df = pd.DataFrame({"user": pd.Series(users), "item": pd.Series(items)}, dtype=np.float32)
evaluator = Evaluator(use_multi_threading=False)
tensor = TensorCoFi.get_model_from_cache()
tfm_tensor = PyTensorCoFi()
tfm_tensor.data_map = tensor.data_map
tfm_tensor.users_size = lambda: tensor.users_size()
tfm_tensor.items_size = lambda: tensor.items_size()
tfm_tensor.get_score = lambda user, item: \
np.dot(tfm_tensor.factors[0][tfm_tensor.data_map[tfm_tensor.get_user_column()][user]],
tfm_tensor.factors[1][tfm_tensor.data_map[tfm_tensor.get_item_column()][item]].transpose())
tfm_tensor.train(data.transpose())
items = df.item.unique()
t = evaluator.evaluate_model(tensor, df, all_items=items, non_relevant_count=100)
tfm = evaluator.evaluate_model(tfm_tensor, df, all_items=items, non_relevant_count=100)
assert abs(t[0] - tfm[0]) < 0.15, \
"Difference between testfm implementation and frappe is to high (%f, %f)" % (t[0], tfm[0])
os.environ.setdefault("DJANGO_SETTINGS_MODULE", "recommendation.default_settings")
from django.core.wsgi import get_wsgi_application
application = get_wsgi_application()
from django.conf import settings
from recommendation.models import Item, User, TensorCoFi, Popularity
# Load user and items
Item.load_to_cache()
User.load_to_cache()
# Load main models
Popularity.load_to_cache()
TensorCoFi.load_to_cache()
if "recommendation.language" in settings.INSTALLED_APPS:
from recommendation.language.models import Locale, Region
Locale.load_to_cache()
Region.load_to_cache()
#if "recommendation.simple_logging" in recommendation.settings.INSTALLED_APPS:
# print("Loading logs to cache...")
# from recommendation.simple_logging.models import LogEntry
# LogEntry.load_to_cache()
# print("done!")
if "recommendation.diversity" in settings.INSTALLED_APPS:
"recommendation.default_settings")
from django.core.wsgi import get_wsgi_application
application = get_wsgi_application()
from django.conf import settings
from recommendation.models import Item, User, TensorCoFi, Popularity
# Load user and items
Item.load_to_cache()
User.load_to_cache()
# Load main models
Popularity.load_to_cache()
TensorCoFi.load_to_cache()
if "recommendation.language" in settings.INSTALLED_APPS:
from recommendation.language.models import Locale, Region
Locale.load_to_cache()
Region.load_to_cache()
#if "recommendation.simple_logging" in recommendation.settings.INSTALLED_APPS:
# print("Loading logs to cache...")
# from recommendation.simple_logging.models import LogEntry
# LogEntry.load_to_cache()
# print("done!")
if "recommendation.diversity" in settings.INSTALLED_APPS:
from recommendation.diversity.models import ItemGenre, Genre
Genre.load_to_cache()