In [50]:
%matplotlib inline
import pandas as pd
from ambry.bundle import BuildBundle
import os
b = BuildBundle(os.getcwd())
for p in b.partitions:
print p.identity
p = b.partitions.all[0]
df = p.pandas
In [51]:
df.head()
Out[51]:
In [53]:
df.reporting_staff_member.value_counts().to_csv('reporters.csv')
In [46]:
df.grade.hist()
Out[46]:
In [24]:
dt = pd.to_datetime(df.date_and_time_of_misbehavior)
In [30]:
df.date_and_time_of_misbehavior.dt.hour.hist(bins=24)
Out[30]:
In [33]:
df.date_and_time_of_misbehavior.dt.month.hist(bins=12)
Out[33]:
In [36]:
(df.first_name + " " + df.last_name).value_counts()
Out[36]:
In [ ]: