numpanprob.py

import numpy as np
import pandas as pd

a = pd.DataFrame({'user_id': [8993,8994,8994,9009,9009],\
                  'amount' : [50.00, 72.00, 72.00, 21.99, 21.99]})

a.set_index('user_id').amount.min()
real_ids = [8993,8994,8994,9009,9009]
fake_ids = [999999,999998]
ids = real_ids+fake_ids
sam = np.random.choice(ids, 15, True)
#sam = filter(lambda x: x in real_ids, sam) #fixes the problem
a.set_index('user_id').ix[sam].dropna(0) 

#whaat? it makes up rows, occasionally with huge values on the order of 1.0e277