#!/usr/bin/env python
# coding: utf-8

# In[54]:


import pandas as pd
import numpy as np


# In[65]:


arr = np.random.randn(100000) * 50
arr


# In[66]:


ref_q1 = pd.DataFrame(arr, columns=["value"]).quantile(0.1).value
print("ref q1 is %s" % ref_q1)


# In[67]:


arr = np.random.permutation(arr)
sub_groups = []
while len(arr) > 0:
    num_rows = len(arr) if len(arr) <= 1000 else np.random.randint(100, 1000)
    print("will take %s rows from %s" % (num_rows, len(arr)))
    sub_group = arr[0:num_rows]
    sub_groups.append(sub_group)
    arr = arr[num_rows:]


# In[68]:


computed = []
for group in sub_groups:
    df = pd.DataFrame(group, columns=["value"])
    computed.append([len(group), df.quantile(0.1).value, df.quantile(0.5).value])

    
df = pd.DataFrame(computed, columns=["population", "q1", "q5"])
q1 = (df.q1*df.population).sum() / df.population.sum()
print("df.population.sum() is %s" % df.population.sum())
print("computed q1 is %s" % q1)


# In[69]:


a = [1,2,3,4,5,6,7]
b = a[0:3]
c = a[3:]
print("b is %s and c is %s" % (b,c))


# In[ ]: