def filter_seen(self, playlist_id, scores):
start_pos = self.URM.indptr[playlist_id]
end_pos = self.URM.indptr[playlist_id + 1]
playlist_profile = self.URM.indices[start_pos:end_pos]
scores[playlist_profile] = -np.inf
return scores
# MARK: - Train and evaluate algorithm
icfrecommendertest = ItemCFKNNRecommender(URM_train, URM_all)
icfrecommendertest.fit(shrink=2, topK=100)
evaluate_algorithm(URM_test, icfrecommendertest, at=10)
icfrecommender = ItemCFKNNRecommender(URM_test, URM_all)
icfrecommender.fit(shrink=2, topK=100)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
for line in target_playlist_file:
start_pos = self.URM.indptr[playlist_id]
end_pos = self.URM.indptr[playlist_id + 1]
playlist_profile = self.URM.indices[start_pos:end_pos]
scores[playlist_profile] = -np.inf
return scores
# MARK: - Train and evaluate algorithm
cbfrecommender = ItemCBFKNNRecommender(URM_train, ICM_album, ICM_artist,
ICM_duration)
cbfrecommender.fit()
evaluate_algorithm(URM_test, cbfrecommender)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
for line in target_playlist_file:
line = line.replace("\n", "")
try:
playlist_id = int(line)
end_pos = self.URM_train.indptr[playlist_id + 1]
playlist_profile = self.URM_train.indices[start_pos:end_pos]
scores[playlist_profile] = -np.inf
return scores
# MARK: - Train and evaluate algorithm
slimbprrecommendertest = SLIM_BPR_Python(URM_train,
positive_threshold=0,
sparse_weights=True)
slimbprrecommendertest.fit(epochs=2, topK=200)
evaluate_algorithm(URM_test, slimbprrecommendertest)
slimbprrecommender = SLIM_BPR_Python(URM_test,
positive_threshold=0,
sparse_weights=True)
slimbprrecommender.fit(epochs=2, topK=200)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
# We are not interested in sorting the popularity value,
# but to order the items according to it
self.popularItems = np.argsort(itemPopularity)
self.popularItems = np.flip(self.popularItems, axis=0)
def recommend(self, user_id, at=10):
recommended_items = self.popularItems[0:at]
return recommended_items
# Evaluate algorithm
print("-- Top Pop Recommender --")
topPopRecommender = TopPopRecommender()
topPopRecommender.fit(URM_train)
evaluate_algorithm(URM_test, topPopRecommender)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
for line in target_playlist_file:
line = line.replace("\n", "")
try:
playlist_id = int(line)
start_pos = self.URM.indptr[playlist_id]
end_pos = self.URM.indptr[playlist_id + 1]
playlist_profile = self.URM.indices[start_pos:end_pos]
scores[playlist_profile] = -np.inf
return scores
# MARK: - Train and evaluate algorithm
hybrid_cf_cbf_recommendertest = Hybrid_CF_CBF_KNNRecommender(
URM_train, URM_all, ICM_album, ICM_artist, ICM_duration)
hybrid_cf_cbf_recommendertest.fit(shrink=3, topK=200)
evaluate_algorithm(URM_test, hybrid_cf_cbf_recommendertest, at=10)
hybrid_cf_cbf_recommender = Hybrid_CF_CBF_KNNRecommender(
URM_test, URM_all, ICM_album, ICM_artist, ICM_duration)
hybrid_cf_cbf_recommender.fit(shrink=3, topK=200)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
class RandomRecommender(object):
def fit(self, URM_train):
self.numItems = URM_train.shape[0]
def recommend(self, user_id, at=10):
recommended_items = np.random.choice(self.numItems, at)
return recommended_items
# Evaluate algorithm
print("-- Random Recommender --")
randomRecommender = RandomRecommender()
randomRecommender.fit(URM_train)
evaluate_algorithm(URM_test, randomRecommender)
# Let's generate recommendations for the target playlists
print("Generating recommendations...")
target_playlist_path = "../input/target_playlists.csv"
target_playlist_file = open(target_playlist_path, 'r')
target_playlist_file.seek(0)
target_playlist_tuples = []
numberOfTargets = 0
for line in target_playlist_file:
line = line.replace("\n", "")
try:
playlist_id = int(line)