def getUsers():
ret = []
with open("data/edinburgh_restaurant_users.json") as f:
data = f.readlines()
data = map(lambda x: x.rstrip(), data)
data_json_str = "[" + ','.join(data) + "]"
data_df = pd.read_json(data_json_str)
for i in range(0, data_df.shape[0]):
datai = data_df.iloc[i]
u = User()
u.setVoteCount(sum(map(int, datai['votes'].values())))
u.setEliteNum(len(datai["elite"]))
for friend in datai["friends"]:
u.addFriend(friend)
u.setID(datai["user_id"])
u.setFanCount(datai["fans"])
u.setFriendCount(len(datai["friends"]))
u.setReviewCount(datai["review_count"])
u.setYelpingSince(datai["yelping_since"])
ret.append(u)
return ret
if fav_id != 0:
break
u.addFavorite(fav_id)
#a long-tailed number of friends
num_friends = int(round(random.lognormvariate(d, .5)))
for i in range(num_friends):
#long-tailed distribution on friends
friend_id = 0
while True:
friend_id = (1 + int(round(random.lognormvariate(4, 5)))) % 10000
if friend_id != 0:
break
u.addFriend(friend_id)
#long-tailed number of purchases
num_purchases = int(round(random.lognormvariate(d / 2, .1))) - 2
for i in range(num_purchases):
pruch_id = 0
while True:
purch_id = (1 + int(round(random.lognormvariate(4, 4.5)))) % 50000
if purch_id != 0:
break
u.addPurchase(purch_id)
#print "%d\t%d\t%d\t%d" %(u.id, len(u.favorites), len(u.friends), len(u.purchases))
print u.toJson()
if fav_id != 0:
break
u.addFavorite(fav_id)
#a long-tailed number of friends
num_friends = int(round(random.lognormvariate(d, .5)))
for i in range(num_friends):
#long-tailed distribution on friends
friend_id = 0
while True:
friend_id = (1+int(round(random.lognormvariate(4, 5))))%10000
if friend_id != 0:
break
u.addFriend(friend_id)
#long-tailed number of purchases
num_purchases = int(round(random.lognormvariate(d/2, .1))) - 2
for i in range(num_purchases):
pruch_id = 0
while True:
purch_id = (1+int(round(random.lognormvariate(4, 4.5))))%50000
if purch_id != 0:
break
u.addPurchase(purch_id)
#print "%d\t%d\t%d\t%d" %(u.id, len(u.favorites), len(u.friends), len(u.purchases))
