# num_eval += 1
is_relevant = np.array([
False, False, False, False, False, False, False, False, False,
False
])
#if True in is_relevant[:3]:
#goodUsers.append(user)
num_eval += 1
cumulative_precision += eval.precision(is_relevant, relevant_items)
cumulative_recall += eval.recall(is_relevant, relevant_items)
cumulative_MAP += eval.MAP(is_relevant, relevant_items)
cumulative_precision /= num_eval
cumulative_recall /= num_eval
cumulative_MAP /= num_eval
result_dict = {
"precision": cumulative_precision,
"recall": cumulative_recall,
"MAP": cumulative_MAP
}
print(result_dict)
with open("../../../Outputs/LightFM_topPop_1_9600_all.csv", 'w') as f:
f.write("user_id,item_list\n")
for user_id in target_users:
f.write(
str(user_id) + "," +
utils.trim(np.array(recommender.recommend(user_id))) + "\n")
outputs[u].append(recommendations.get(key).recommend(u))
elif target_age < 0 and target_region < 0:
if u in users_dict[0]:
outputs[u].append(recommendations.get(0).recommend(u))
else:
outputs[u].append(recommendations.get(-1).recommend(u))
#print(u)
print("DONE!")
with open('../../Dataset/topPopularRecommendations.p', 'wb') as fp:
pickle.dump(outputs, fp, protocol=pickle.HIGHEST_PROTOCOL)
return outputs
#output = create_recommendations()
#exit()
with open('../../Dataset/topPopularRecommendations.p', 'rb') as fp:
output = pickle.load(fp)
f = open("../../Outputs/TopPop_freeze.csv", 'w')
f.write("user_id,item_list\n")
for u in utils.get_target_users("../../Dataset/target_users_freeze.csv",seek=8):
if u in output.keys():
if len(output[u])==1:
f.write(str(u) + "," + utils.trim(output[u][0]) + "\n")
else:
#print('user want MIXER')
f.write(str(u) + "," + utils.trim(select1(output[u])) + "\n")
else:
print("WE HAVE A PROBLEM {0} IS NOT HERE".format(u))
}
return result_dict
URM_test = sps.csr_matrix(sps.load_npz("../Dataset/URM/data_test.npz"))
#URM_val = sps.csr_matrix(sps.load_npz("../Dataset/old/data_validation.npz"))
targetUsers = util.get_target_users("../Dataset/target_users.csv", seek=9)
print("TESTING")
res_test = run(targetUsers, URM_test)
print(res_test["MAP"])
"""
print(res_test)
print("VALIDATION")
res_val = run(targetUsers, URM_val)
print(res_val["MAP"])
#print(res_val)
print((res_val["MAP"]+res_test["MAP"])/2)
"""
with open("../Outputs/hybrid_norm_15_1_1.csv", 'w') as f:
f.write("user_id,item_list\n")
for user_id in targetUsers:
#print(user_id)
f.write(
str(user_id) + "," +
util.trim(np.array(res_test["RecSys"][user_id])) + "\n")
util.compare_csv("../Outputs/truth.csv", "../Outputs/hybrid_norm_15_1_1.csv")
score[j] += 1
else:
score[j] = 1
max = 0
#print(score)
rec = list()
rec.append(list())
for w in sorted(score, key=score.get, reverse=True):
if max > 9:
break
rec[0].append(w)
max += 1
count += 1
#print("user {0} in more than 1 cluster".format(u))
#print(rec)
f.write(str(u) + "," + utils.trim(np.array(rec[0])) + "\n")
print(count)
"""
pyplot.plot(MAP_pureSVD_per_group, label="pureSVD")
pyplot.plot(MAP_sslim_per_group, label="sslim")
pyplot.plot(MAP_CFItem_per_group, label="CFItem")
pyplot.plot(MAP_slim_per_group, label="slim")
pyplot.plot(MAP_CFUser_per_group, label="CFUser")
pyplot.plot(MAP_CBItem_per_group, label="CBItem")
pyplot.plot(MAP_P3a_per_group, label="P3a")
pyplot.plot(MAP_P3b_per_group, label="P3b")
pyplot.plot(MAP_NMF_per_group, label="elasticNet")
pyplot.ylabel('MAP')
pyplot.xlabel('User Group')
scores[user_profile] = -np.inf
return scores
itemColl = ItemKNNCFRecommender(URM_all)
itemColl.fit(shrink=50, topK=10)
elasticNet = SLIMElasticNetRecommender(URM_all)
elasticNet.fit()
users = utils.get_target_users("../../Dataset/target_users.csv")
hybridrecommender = ItemKNNScoresHybridRecommender(URM_all, itemColl, elasticNet)
hybridrecommender.fit(0.3)
with open("../../../Outputs/itemColl+elasticNet_0.3.csv", 'w') as f:
f.write("user_id,item_list\n")
for user_id in users:
f.write(str(user_id) + ", " + utils.trim(hybridrecommender.recommend(user_id)[:10]) + "\n")
hybridrecommender.fit(0.5)
with open("../../../Outputs/itemColl+elasticNet_0.5.csv", 'w') as f:
f.write("user_id,item_list\n")
for user_id in users:
f.write(str(user_id) + ", " + utils.trim(hybridrecommender.recommend(user_id)[:10]) + "\n")
hybridrecommender.fit(0.7)
with open("../../../Outputs/itemColl+elasticNet_0.7.csv", 'w') as f:
f.write("user_id,item_list\n")
for user_id in users:
f.write(str(user_id) + ", " + utils.trim(hybridrecommender.recommend(user_id)[:10]) + "\n")
l=2.299,
n=True)
with open("../../Outputs/hybrid_norm_final2.csv", 'w') as f:
f.write("user_id,item_list\n")
user_batch_start = 0
while user_batch_start < len(users):
user_batch_end = user_batch_start + 1000
user_batch_end = min(user_batch_end, len(users))
test_user_batch_array = np.array(
users[user_batch_start:user_batch_end])
user_batch_start = user_batch_end
# Compute predictions for a batch of users using vectorization, much more efficient than computing it one at a time
recommended_items_batch_list, scores_batch = rec_sys.recommend(
test_user_batch_array,
remove_seen_flag=True,
cutoff=10,
remove_top_pop_flag=False,
remove_custom_items_flag=False,
return_scores=True)
i = 0
for user_id in test_user_batch_array:
# print(user_id)
f.write(
str(user_id) + "," +
utils.trim(np.array(recommended_items_batch_list[i])) + "\n")
i = i + 1