def register(userid,password,email):
new_user=user_info()
new_user.userid=userid
new_user.password=password
new_user.email=email
make_new_user(new_user)
username_list.append(userid)
password_list.append(password)
email_list.append(email)
#Now create bandits
#total_bandits = range(0,k)
movie_dict = read_genre() # This gives us the actions for each of the bandits. Here the actions are movies specific to a genre
b_keys = movie_dict.keys()
bandit_dict = {}
bandits = [] # list of bandit objects
for b in b_keys:
acts = movie_dict[b]
bandit = Bandit(b,acts)
bandit.set_count()
bandit_dict[b] = bandit
# Getting the point class information
points = user_info('users.dat',bandit_dict)
#print points[0]
#print points
point_values = [x.value for x in points]
#Set the Q values for each of the points
for b in bandit_dict.values():
for p in points:
b.set_Q(p)
# Initialize some values
# sample some users for testing and recommendation
from collections import defaultdict
from aff_clustering import *
from user_info import *
#Load the user and movie ratings dictionaries using pickle
user_rating_dict = pickle.load(open('user_rating_dict'))
movie_rating_dict = pickle.load(open('movie_rating_dict'))
# Now define the number of clusters ; it will be equal to the number of movies
no_of_movies = len(movie_rating_dict.keys())
k = 18 # Currently set to the number of genres
#Now let us perform one round of clustering and display the results
print 'The number of clusters is:' + str(k) + '\n'
points = user_info('users.dat')
clusters = kmeans(points,k)
#print clusters
'''
Initializing the affinity values for each of the data points.
Affinity values are stochastic for each of the data points. So initializing equal affinity scores for all the points
Investigate: Could it be a fuzzy set?
'''
init_affin = 1/k
aff_dict = defaultdict(list)
aff_list = []
for i in range(k):
aff_list.append(init_affin)
