In [107]:

    

import pandas as pd
import seaborn as sns
from sklearn.linear_model import RidgeCV
import numpy as np


    

In [52]:

    

data_path = '../../data/cleaned/UCB_dept_merge.csv'


    

In [56]:

    

date_columns = ['po_closed_date', 'creation_date']


    

In [57]:

    

data = pd.read_csv(data_path, parse_dates=date_columns)


    

In [25]:

    

%matplotlib inline


    

In [115]:

    

count_cutoff = 1000
count_name = 'department_name'


    

In [116]:

    

counts = data.groupby(count_name).count().iloc[:, 0].\
    sort(ascending=False, inplace=False)
counts[counts > count_cutoff].plot(kind='bar', figsize=(10, 5))


    

    
Out[116]:





<matplotlib.axes._subplots.AxesSubplot at 0x131c4ead0>






    

In [45]:

    

counts[counts > count_cutoff].plot(kind='pie')


    

    
Out[45]:





<matplotlib.axes._subplots.AxesSubplot at 0x13638de50>






    

In [98]:

    

ndays = [i.days for i in data.days_to_close if not pd.isnull(i)]


    

In [101]:

    

data['days_to_close'] = data['po_closed_date'] - data['creation_date']
data['days_to_close'] = data['days_to_close'].astype('timedelta64[D]')


    

    




CPU times: user 37.6 ms, sys: 15.5 ms, total: 53.2 ms
Wall time: 50.9 ms

In [121]:

    

use_depts = counts[counts > 2000].index.values


    

In [155]:

    

_ = data.query('department_name in @use_depts').\
    boxplot(column='days_to_close', by='department_name', rot=90)


    

In [104]:

    

data.days_to_close.hist(bins=100)


    

    
Out[104]:





<matplotlib.axes._subplots.AxesSubplot at 0x131795ad0>