def user_clustering():
users = []
with open(FILE_DIR + file_path[0]) as f:
for line in f:
_, _, _, user_data, _ = InputData.split_data(line)
if user_data not in users: users.append(user_data)
if len(users) % 10000 == 0: print(len(users))
print(len(users))
kmeans = UserCluster(N_CLUSTERING)
kmeans.fit(features)
joblib.dump(kmeans, 'kmeans.pkl')
return kmeans
def user_clustering():
users = []
with open('../analytics/stdev/usercluster200.csv') as f:
for line in f:
_, _, _, user_data, _ = InputData.split_data(line)
if user_data not in users: users.append(user_data)
if len(users) % 10000 == 0: print(len(users))
print(len(users))
kmeans = UserCluster(N_CLUSTERING)
kmeans.fit(features)
joblib.dump(kmeans, 'kmeans.pkl')
return kmeans
def run_enviroment(algorithms, cluster_model):
ite = 0
for file_name in file_path:
with open(FILE_DIR + file_name) as f:
old = {}
old['LinearedUCB'] = 0
old['OriginalUCB'] = 0
old_imp = {}
old_imp['LinearedUCB'] = 0
old_imp['OriginalUCB'] = 0
click_count = 0
click_eq_count = 0
ctr_list = {}
ctr_list['LinearedUCB'] = []
ctr_list['OriginalUCB'] = []
for line in f:
_, click_article_id, click, user_data, article_pool = InputData.split_data(
line)
click_count += click
userID = cluster_model.predict_cluster(user_data)[0]
for name, alg in algorithms.items():
decide_id = alg.decide(userID, user_data, article_pool)
if evaluate(click_article_id, decide_id, click, name, ite,
line):
# if click == 1: print(click, algorithms['OriginalUCB'].get_prob_check(userID, article_pool[decide_id]))
ctr_list[name].append((reward[name] - old[name]) /
(count[name] - old_imp[name]))
alg.update(userID, user_data, article_pool[decide_id],
click)
if count[name] % 2000 == 0:
# print(ite, name, reward[name], count[name], reward[name]/count[name], reward[name]-old[name], count[name]-old_imp[name], (reward[name]-old[name])/(count[name]-old_imp[name]), click_count)
print(name, np.mean(ctr_list[name]),
reward[name] - old[name],
count[name] - old_imp[name], reward[name],
count[name])
old[name] = reward[name]
old_imp[name] = count[name]
click_count = 0
click_eq_count = 0
ctr_list[name] = []
count[name] += 1
# alg.save_weight(name + '_weight_' + str(N_CLUSTERING) + '_3.csv')
ite += 1
return
def run_enviroment(algorithms, cluster_model):
iteration = 0
ite = {}
click_count = {}
old = {}
old2 = {}
old_imp = {}
old_imp2 = {}
ctr = {}
ctr_list = {}
rew = {}
writer = {}
now = datetime.datetime.today()
for name in algorithms.keys():
old[name] = 0
old_imp[name] = 0
old2[name] = 0
old_imp2[name] = 0
ctr_list[name] = [1.]
ctr[name] = []
rew[name] = [0]
ite[name] = 0
click_count[name] = 1
# writer[name] = open('../data/'+name+'_experiment_log_'+str(now)+'.csv', 'w')
for file_name in file_path:
with open(FILE_DIR + file_name) as f:
for line in f:
_, click_article_id, click, user_data, article_pool = InputData.split_data(
line)
article_pool = article_pool_10(article_pool, click_article_id)
userID = cluster_model.predict_cluster(user_data)[0]
# user_item_log[userID][]
for name, alg in algorithms.items():
ite[name] += 1
click_count[name] += click
decide_id = alg.decide_try(userID, user_data, article_pool)
if name == 'OriginalUCB' and iteration < 150000 and random.random(
) < (1. - (iteration * 0.5 / 150000)):
decide_id = random.choice(list(article_pool.keys()))
random_ctr = click_count[name] / ite[name]
cumulated_ctr = reward[name] / count[name]
relative_ctr = cumulated_ctr / random_ctr
if iteration % 2000 == 0:
ctr[name].append(np.mean(ctr_list[name]))
#writer[name].write(str(float('{:.5f}'.format(np.mean(ctr[name]))))+','+str(float('{:.5f}'.format(np.mean(ctr_list[name]))))+','+str(float('{:.5f}'.format(cumulated_ctr)))+'\n')
if iteration % 10000 == 0:
print(
iteration, name,
float('{:.5f}'.format(np.mean(ctr[name]))),
float('{:.5f}'.format(np.mean(
ctr_list[name]))),
reward[name] - old2[name],
count[name] - old_imp2[name], reward[name],
count[name],
float('{:.5f}'.format(cumulated_ctr)),
float('{:.5f}'.format(relative_ctr)))
old2[name] = reward[name]
old_imp2[name] = count[name]
rew[name].append(reward[name] - old[name])
old[name] = copy.deepcopy(reward[name])
old_imp[name] = copy.deepcopy(count[name])
ctr_list[name] = [1.]
count[name] += 1
# alg.save_weight(name + '_weight_' + str(N_CLUSTERING) + '_3.csv')
if evaluate(click_article_id, decide_id, click, name, ite,
line):
ctr_list[name].append(relative_ctr)
alg.update(userID, user_data, article_pool[decide_id],
click, decide_id)
if name == 'CLUB':
n_components = alg.updateGraphClusters(
userID, 'False')
iteration += 1
gragh(ctr, rew)
for name, alg in algorithms.items():
print(name, np.mean(ctr[name]))
# writer[name].close()
alg.memory_item_num()
return
def predict_cluster(self, feature):
return self.model.predict(feature)
def model_load(self, model_file):
return joblib.load('/Users/chan-p/GitHub/ReserchBandit/system/model/' +
model_file)
if __name__ == '__main__':
n = 1000
km = UserCluster(n)
users = []
with open('../../../Desktop/R6/ydata-fp-td-clicks-v1_0.20090501') as f:
for line in f:
timestamp, click_article_id, click, user_data, article_pool = InputData.split_data(
line)
users.append(user_data)
if len(users) == 3000000:
km.fit(users)
joblib.dump(
km,
'./model/model' + str(n) + '_' + str(len(users)) + '.pkl')
# k = km.model_load('model40.pkl')
'''
for user in users:
print(k.predict_cluster(user))
'''
am = UserCluster(n)
am.fit(users)
joblib.dump(am,
'./model/model' + str(n) + '_' + str(len(users)) + '.pkl')