def compute_conditional_probabilities():
logger.debug("Computing conditional probabilities.")
user_watches = users.get_user_watches()
repo_frequencies = users.get_repo_frequencies()
# Prune watchlists to only those w/ greater than 1 watch
watches_list = [w for w in user_watches.values() if len(w) > 1]
watches_size = len(watches_list)
logger.debug("Watches size {0}".format(watches_size))
cprob = collections.defaultdict(dict)
count = 0
for watches in watches_list:
count += 1
logger.debug("Processing watch {0} of {1}".format(count, watches_size))
for i in watches:
for j in watches:
if i == j: continue
if not j in cprob[i]:
cprob[i][j] = 1, 1/repo_frequencies[i]
else:
cofreq = cprob[i][j][0] + 1
cprob[i][j] = cofreq, cofreq/repo_frequencies[i]
persist_conditional_probabilities(cprob)
def __init__(self):
self.user_watches = users.get_user_watches()
self.repo_freqs = users.get_repo_frequencies()
self.repos = repos.get_repos()
self.test_users = users.get_test_user_ids()