def main(args):
start_time = time.time()
users, repos, popular_repos, superprojects = load_data(args)
print "Processing test users"
test = open(args[3], 'r')
results = open(args[4], 'w')
for line in test.readlines():
user_id = int(line.strip())
if not user_id in users:
user = User(user_id)
else:
user = users[user_id]
suggested_repos = suggest_repos(repos, popular_repos, users, user,
superprojects)
suggested_repos.sort()
results.write('%d:' % user.id)
results.write(','.join([str(x.id) for x in suggested_repos]))
results.write('\n')
results.flush()
test.close()
results.close()
print "Done in %d seconds" % (time.time() - start_time)
def main(args):
start_time = time.time()
print "Loading results"
target = load_data(args[0])
results = load_data(args[1])
total = len(target)
missed = target.difference(results)
guessed = total - len(missed)
print "%d/%d, %d%%" % (guessed, len(target),
(guessed * 100.0) / len(target))
print "Done in %d seconds" % (time.time() - start_time)
def main(args):
start_time = time.time()
users, repos, popular_repos = load_data(args)
print "Unique users: %d" % len(users)
print "Unique repos: %d" % len(repos)
print "Done in %d seconds" % (time.time() - start_time)
def main(args):
start_time = time.time()
users, repos, popular_repos = load_data(args)
target_data = remove_target_data(users)
print "Writing test data"
# XXX create directory
output_sample("%s/data.txt" % args[3], users)
output_test("%s/test.txt" % args[3], target_data)
output_target("%s/target.txt" % args[3], target_data)
print "Done in %d seconds" % (time.time() - start_time)