def load_top_recommendations(self, n_recommendations, predictions,
test_data, fold):
"""
This method loads the top n recommendations into a local variable.
:param int n_recommendations: number of recommendations to be generated.
:param int[][] predictions: predictions matrix (only 0s or 1s)
:returns: A matrix of top recommendations for each user.
:rtype: int[][]
"""
for user in range(self.ratings.shape[0]):
# Bug discovered by Mostafa
#nonzeros = self.test_indices[(user * (1 + fold))]
#Fix:
nonzeros = self.test_indices[user * self.k_folds + fold]
top_recommendations = TopRecommendations(n_recommendations)
for index in nonzeros:
index = int(index)
top_recommendations.insert(index, predictions[user][index])
self.recommendation_indices[user] = list(
reversed(top_recommendations.get_indices()))
top_recommendations = None
self.recs_loaded = True
return self.recommendation_indices
def recommend_items(self, user_id, num_recommendations=10):
"""
Get recommendations for a user. Based on the predictions returned by get_predictions
:param int user_id: The id of the user.
:param int num_recommendations: The number of recommended items.
:returns:
A zipped object containing list of tuples; first index is the id of the document
and the second is the value of the calculated recommendation.
:rtype: zip
"""
top_recommendations = TopRecommendations(num_recommendations)
user_ratings = self.get_predictions()[user_id]
for i in range(len(user_ratings)):
top_recommendations.insert(i, user_ratings[i])
return zip(top_recommendations.get_indices(), top_recommendations.get_values())
def load_top_recommendations(self, n_recommendations, predictions, test_data):
"""
This method loads the top n recommendations into a local variable.
:param int n_recommendations: number of recommendations to be generated.
:param int[][] predictions: predictions matrix (only 0s or 1s)
"""
for user in range(self.ratings.shape[0]):
nonzeros = test_data[user].nonzero()[0]
top_recommendations = TopRecommendations(n_recommendations)
for index in nonzeros:
top_recommendations.insert(index, predictions[user][index])
self.recommendation_indices[user] = list(reversed(top_recommendations.get_indices()))
top_recommendations = None
self.recs_loaded = True