In [365]:

import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import pandas as pd
from __future__ import division
import plotly.offline
import plotly.plotly as ply
import plotly.figure_factory as ff  # generate table
from plotly.graph_objs import *     # generate bar charts etc.

from plotly.offline import download_plotlyjs, init_notebook_mode, plot, iplot

%matplotlib inline

init_notebook_mode(connected=True)

Data from MotherJones¶

US Mass Shootings, 1982-2018: Data From Mother Jones’ Investigation --The full data set from our in-depth investigation into mass shootings.

by Mark Follman, Gavin Aronsen and Deanna Pan

In [366]:

fname='MotherJones_MassShooting_1982-2018.csv'
df = pd.read_csv(fname,converters={"Fatalities":int})

# correct a few indices
df.loc[df['Venue'] == '\nWorkplace', 'Venue'] = 'Workplace'
df.loc[df['Venue'] == 'Other\n', 'Venue'] = 'Other'
df.groupby(['Venue'])['Fatalities'].agg({'Fatalities': ['sum']})

Out[366]:

	Fatalities
	sum
Venue
Airport	5
Military	38
Other	381
Religious	57
School	162
Workplace	173

In [367]:

#df.loc[df['Venue'] == 'Other']['Sources'][2]
sum_per_venue.index

Out[367]:

Index([u'Airport', u'Military', u'Other', u'Religious', u'School',
       u'Workplace'],
      dtype='object', name=u'Venue')

In [368]:

sum_per_venue=df.groupby(['Venue'])['Fatalities'].agg(['sum'])
count_per_venue=df.groupby(['Venue']).agg('count')['Case']

In [369]:

sum_per_venue['count'] = count_per_venue
sum_per_venue['mean'] = np.around(sum_per_venue['sum']/count_per_venue,decimals=1)
sum_per_venue['Venue'] = sum_per_venue.index
sum_per_venue = sum_per_venue.reindex(columns=['Venue','sum','count','mean'])
table = ff.create_table(sum_per_venue,height_constant=30)
table.layout.width=700
#ply.iplot(table, filename='table1')
iplot(table, filename='table1')

In [371]:

#sum_per_venue.iloc[:,2].plot(kind='bar')
df_mj = sum_per_venue.drop(['Other'])

ctable = ['rgb(145,191,219)']*5
ctable[3] = 'rgba(222,45,38,0.8)'
width = list(np.full(5,0.5))

data = [Bar(x=df_mj['Venue'],
            y=df_mj['mean'],marker=dict(color=ctable),width=width)]

layout = Layout(autosize=False,width=700,height=500, title='Mean # of Fatalities per Incident')
         #margin=Margin(l=20,r=20,b=20,t=20,pad=4),
         #paper_bgcolor='#7f7f7f',
         #plot_bgcolor='#c7c7c7')
fig = Figure(data=data,layout=layout)
#ply.iplot(fig, filename='mean_fatalities_per_incidents')
iplot(fig, filename='mean_fatalities_per_incidents')

In [372]:

#sum_per_venue.plot.barh(y=['sum','count'],stacked='True',colormap='jet')
ctable1 = ['rgb(145,191,219)']*5
ctable1[3] = 'rgba(210,45,38,0.8)'
ctable2 = ['rgb(190,191,219)']*5
ctable2[3] = 'rgba(250,45,38,0.8)'
width = list(np.full(5,0.4))

data1 = Bar(x=df_mj['Venue'],
            y=df_mj['sum'],marker=dict(color=ctable1),width=width,name='Sum')
data2 = Bar(x=df_mj['Venue'],
            y=df_mj['count'],marker=dict(color=ctable2),width=width,name='Count')

layout = Layout(autosize=False,width=700,height=500, title='# of Mass Shootings and Total # of Fatalities')
         #margin=Margin(l=20,r=20,b=20,t=20,pad=4),
         #paper_bgcolor='#7f7f7f',
         #plot_bgcolor='#c7c7c7')
data=[data2,data1]
fig = Figure(data=data,layout=layout)
#ply.iplot(fig, filename='count_sum_fatalities_by_venue')
iplot(fig, filename='count_sum_fatalities_by_venue')

CDC data¶

In [373]:

fname='CDC_MassShooting_2009-2016.csv'
df_cdc=pd.read_csv(fname)

In [374]:

df_cdc_school=df_cdc.groupby('School')['Number Killed [CALCULATED]'].agg(['count','sum'])
df_cdc_school

Out[374]:

	count	sum
School
No	150	783
Yes	6	65

cross check between these two datasets¶

Binghamton,NY 04-03-2009 mass shooting is listed as a 'School' incident, but not in Mother Jones'
Santa Monica 6/7/2013 is missing from MotherJones
The rest are all covered in Mother Jones' report.

In [375]:

print(len(df[df['Venue']=='School']))
print(len(df_cdc[df_cdc['School']=='Yes']))

16
6

Combine CDC and MotherJones¶

In [376]:

df_cdc[df_cdc['School']=='Yes']

Out[376]:

	Date	City	State	Listed in FBI SHR (2009-2012)	TOTAL SHOT (Not Including Shooter)	Number Killed [CALCULATED]	Number Injured	Females Killed	Males Killed	Children Killed (17 and Under)	...	School	Workplace	Multiple	Other (Note)	Solely Home But Not DV [CALCULATED]	Red Flag	Subsequent FBI Terrorism Investigation	Public Space	Gun-Free Zone	Took place exclusively in private residence(s)
9	4/3/09	Binghamton	NY	Yes	17	13	4	11	2	0	...	Yes	No	No	NaN	No	No	No	Yes	Yes	no
65	4/2/12	Oakland	CA	Yes	7	7	0	6	1	0	...	Yes	No	No	NaN	No	No	No	Yes	Yes	no
79	12/14/12	Newtown	CT	No	29	27	2	19	8	20	...	Yes	No	No	NaN	No	No	No	Yes	Yes	no
89	6/7/13	Santa Monica	CA	NaN	9	5	4	2	3	0	...	Yes	No	Yes	home, road, school	No	No	No	Yes	Yes	no
113	10/24/14	Marysville	WA	NaN	5	4	1	3	1	4	...	Yes	No	No	NaN	No	No	No	Yes	Yes	no
136	10/1/15	Roseberg	OR	NaN	18	9	9	3	6	0	...	Yes	No	No	NaN	No	No	No	Yes	No	no

6 rows × 49 columns

In [377]:

binary_df = pd.DataFrame(df_cdc_school)
binary_df.loc['Yes']=sum_per_venue.iloc[4]  
binary_df['mean'] = binary_df['sum']/binary_df['count']

binary_df.round(1)

Out[377]:

	count	sum	mean
School
No	150	783	5.2
Yes	16	162	10.1

In [378]:

ctable = ['rgb(145,191,219)','rgba(222,45,38,0.8)']
width = list(np.full(2,0.5))

data = [Bar(y=['Not School','School'],x=binary_df['mean'],orientation='h',marker=dict(color=ctable),width=width)]

layout = Layout(autosize=False,width=700,height=500, title='Mean # of Fatalities per Incident: School vs. Not School')
         #margin=Margin(l=20,r=20,b=20,t=20,pad=4),
         #paper_bgcolor='#7f7f7f',
         #plot_bgcolor='#c7c7c7')
fig = Figure(data=data,layout=layout)
#ply.iplot(fig, filename='mean_fatalities_school_no_school')
iplot(fig, filename='mean_fatalities_school_no_school')

In [ ]: