Exploratory analysis of London street level crime 2015-2017
- 16 mins
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
plt.style.use('ggplot')
%matplotlib inline
Lets load the dataset
data source : https://data.police.uk/data/archive/
data = pd.read_csv('crime.csv')
check the number of rows in the dataset
data.shape
(18000, 13)
the dataset has 18,000 rows and 13 columns
lets check the first five rows
data.iloc[:5,:]
| Unnamed: 0 | Crime ID | Month | Reported by | Falls within | Longitude | Latitude | Location | LSOA code | LSOA name | Crime type | Last outcome category | Context | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 80e07583f4bd74b85e457d92eef5d014e4e8d7b0eab0dc... | 2015-01 | City of London Police | City of London Police | -0.106453 | 51.518207 | On or near Charterhouse Street | E01000916 | Camden 027B | Bicycle theft | Unable to prosecute suspect | NaN |
| 1 | 1 | 6589894ebc515f501527628eb650d52a6f031116eb0ada... | 2015-01 | City of London Police | City of London Police | -0.111497 | 51.518226 | On or near Pedestrian Subway | E01000914 | Camden 028B | Burglary | Investigation complete; no suspect identified | NaN |
| 2 | 2 | e6dc6a4a33ed886c7c72beaff0c5de92cc35cd2f76c6e5... | 2015-01 | City of London Police | City of London Police | -0.111497 | 51.518226 | On or near Pedestrian Subway | E01000914 | Camden 028B | Burglary | Unable to prosecute suspect | NaN |
| 3 | 3 | b6e6462d45d0d7f4258d57628cab4c8988dc41ac675b63... | 2015-01 | City of London Police | City of London Police | -0.111497 | 51.518226 | On or near Pedestrian Subway | E01000914 | Camden 028B | Other theft | Investigation complete; no suspect identified | NaN |
| 4 | 4 | 769e1aa86e62b5f3c4c08c8c140147a275ca721d0801ba... | 2015-01 | City of London Police | City of London Police | -0.113767 | 51.517372 | On or near Stone Buildings | E01000914 | Camden 028B | Theft from the person | Investigation complete; no suspect identified | NaN |
lets remove the crime Id and unamed 0 columns
del data['Crime ID']
del data['Unnamed: 0']
data.head(3)
| Month | Reported by | Falls within | Longitude | Latitude | Location | LSOA code | LSOA name | Crime type | Last outcome category | Context | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 2015-01 | City of London Police | City of London Police | -0.106453 | 51.518207 | On or near Charterhouse Street | E01000916 | Camden 027B | Bicycle theft | Unable to prosecute suspect | NaN |
| 1 | 2015-01 | City of London Police | City of London Police | -0.111497 | 51.518226 | On or near Pedestrian Subway | E01000914 | Camden 028B | Burglary | Investigation complete; no suspect identified | NaN |
| 2 | 2015-01 | City of London Police | City of London Police | -0.111497 | 51.518226 | On or near Pedestrian Subway | E01000914 | Camden 028B | Burglary | Unable to prosecute suspect | NaN |
lets check the month datatype
data.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 18000 entries, 0 to 17999
Data columns (total 11 columns):
Month 18000 non-null object
Reported by 18000 non-null object
Falls within 18000 non-null object
Longitude 17129 non-null float64
Latitude 17129 non-null float64
Location 18000 non-null object
LSOA code 17129 non-null object
LSOA name 17129 non-null object
Crime type 18000 non-null object
Last outcome category 14801 non-null object
Context 0 non-null float64
dtypes: float64(3), object(8)
memory usage: 984.4+ KB
it appears the month datatype is object, we need to convert it to datetime to be able to manipulate it
import datetime
data['Month'] = pd.to_datetime(data['Month'],yearfirst=True)
data.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 18000 entries, 0 to 17999
Data columns (total 11 columns):
Month 18000 non-null datetime64[ns]
Reported by 18000 non-null object
Falls within 18000 non-null object
Longitude 17129 non-null float64
Latitude 17129 non-null float64
Location 18000 non-null object
LSOA code 17129 non-null object
LSOA name 17129 non-null object
Crime type 18000 non-null object
Last outcome category 14801 non-null object
Context 0 non-null float64
dtypes: datetime64[ns](1), float64(3), object(7)
memory usage: 1.0+ MB
lest check location based on incicence of crime
data.Location.value_counts()
No Location 871
On or near Police Station 607
On or near Pedestrian Subway 607
On or near Supermarket 532
On or near Parking Area 487
On or near Great St Helen'S 354
On or near Conference/Exhibition Centre 318
On or near Nightclub 316
On or near Blomfield Street 304
On or near Queen Victoria Street 297
On or near St Martin'S Le Grand 296
On or near Bride Lane 296
On or near New Change 284
On or near Shopping Area 274
On or near Fish Street Hill 269
On or near Clement'S Lane 235
On or near Gracechurch Street 204
On or near Fleet Street 194
On or near Artillery Lane 172
On or near Leadenhall Street 166
On or near Fetter Lane 158
On or near Finch Lane 149
On or near Philpot Lane 149
On or near Bow Lane 141
On or near Cheapside 139
On or near Bell Inn Yard 138
On or near Bishopsgate 133
On or near Bear Alley 132
On or near Primrose Street 130
On or near Eastcheap 129
...
On or near St Dunstan'S Lane 3
On or near Mount Pleasant 3
On or near Old Buildings 2
On or near Dysart Street 2
On or near Amen Court 2
On or near Sandy Lane 2
On or near Tooley Street 2
On or near Shadwell Gardens 2
On or near Goswell Place 2
On or near Mark Street 1
On or near Milk Street 1
On or near Little Essex Street 1
On or near Farringdon Street 1
On or near Cripplegate Street 1
On or near Timber Street 1
On or near Goldsmith Street 1
On or near Stoney Lane 1
On or near Fournier Street 1
On or near Nelson Terrace 1
On or near Whiskin Street 1
On or near Queen Street Place 1
On or near Tower Royal 1
On or near Mucking Wharf Road 1
On or near Folgate Street 1
On or near Holywell Row 1
On or near Banner Street 1
On or near Upper Ground 1
On or near Haven Quays 1
On or near Weaver'S Lane 1
On or near Bartlett Court 1
Name: Location, Length: 335, dtype: int64
lets visualize top ten crime location
def adjust_plot(ax):
plt.rc('xtick',labelsize=22)
plt.rc('ytick',labelsize=22)
# create a list to collect the plt.patches data
totals = []
# find the values and append to list
for i in ax.patches:
totals.append(i.get_height())
# set individual bar lables using above list
total = sum(totals)
# set individual bar lables using above list
for i in ax.patches:
# get_x pulls left or right; get_height pushes up or down
ax.text(i.get_x()-.03, i.get_height()+.5, \
str(round((i.get_height()/total)*100, 2))+'%', fontsize=15,
color='black')
ax = data.Location.value_counts().head(10).plot(kind='bar',figsize=(15,6),title='Crime Location in London city')
plt.title("Crime Location in London 2015-2017", fontname='Ubuntu', fontsize=18,
fontstyle='italic', fontweight='bold',color='green')
adjust_plot()
it turned out location was unknown for 871 times.Aside from No location, most crime happnede very close to police station!!!
lets visualize the 5 least crime location
data.Location.value_counts().tail(5).plot(kind='bar',figsize=(13,5))
<matplotlib.axes._subplots.AxesSubplot at 0xacba93ec>
we found that the most crime free location are Banner Street,Upper Ground,Weaver’s Lane
lets check crime type
data['Crime type'].value_counts()
Other theft 4081
Anti-social behaviour 3199
Violence and sexual offences 2536
Shoplifting 1911
Theft from the person 1270
Drugs 996
Bicycle theft 934
Burglary 665
Public order 661
Criminal damage and arson 659
Other crime 470
Vehicle crime 402
Possession of weapons 111
Robbery 105
Name: Crime type, dtype: int64
other theft, Anti Social behaviour and Sexual behavoiurs is the most prevalent
ax = data['Crime type'].value_counts().plot(kind='bar',figsize=(15,6),title='Crime type in Londo--2015-2017')
plt.title("Crime Type in London 2015-2017", fontname='Ubuntu', fontsize=18,
fontstyle='italic', fontweight='bold',color='green')
adjust_plot(ax)
Location in which Violence and Sexual offences occur the most
data[data['Crime type'] == 'Violence and sexual offences']['Location'].value_counts()
No Location 178
On or near Great St Helen'S 104
On or near Police Station 101
On or near Supermarket 78
On or near Pedestrian Subway 75
On or near Nightclub 65
On or near Parking Area 64
On or near Conference/Exhibition Centre 55
On or near Blomfield Street 53
On or near Queen Victoria Street 51
On or near Leadenhall Street 50
On or near Fish Street Hill 43
On or near Shopping Area 34
On or near Philpot Lane 33
On or near Moorgate 32
On or near Wormwood Street 30
On or near Gracechurch Street 30
On or near St Martin'S Le Grand 29
On or near Finch Lane 29
On or near Bride Lane 29
On or near Watling Court 28
On or near St Swithin'S Lane 27
On or near New Broad Street 25
On or near New Change 23
On or near Wood Street 23
On or near Mark Lane 21
On or near Moor Lane 20
On or near Camomile Street 20
On or near Distaff Lane 20
On or near Creed Lane 19
...
On or near Arthur Street 1
On or near Fournier Street 1
On or near Cloth Street 1
On or near Old Seacoal Lane 1
On or near Lloyd'S Avenue 1
On or near Bell Yard 1
On or near Sandy'S Row 1
On or near Cursitor Street 1
On or near Amen Corner 1
On or near Norton Folgate 1
On or near St Dunstan'S Hill 1
On or near St Mary Axe 1
On or near Copthall Avenue 1
On or near Leadenhall Place 1
On or near Monkwell Square 1
On or near Tudor Street 1
On or near Lombard Lane 1
On or near Portsoken Street 1
On or near Pilgrim Street 1
On or near Nun Court 1
On or near Temple Lane 1
On or near South Place Mews 1
On or near John Carpenter Street 1
On or near Basinghall Street 1
On or near King'S Arms Yard 1
On or near Charterhouse Street 1
On or near Moorgate Place 1
On or near Old Square 1
On or near Billiter Street 1
On or near Charterhouse Mews 1
Name: Location, Length: 256, dtype: int64
sex_crime = data[data['Crime type'] == 'Violence and sexual offences']['Location'].value_counts()
ax = sex_crime.head().plot(kind='bar',figsize=(15,6),title='Location of Violence and Sexual Offence')
plt.title("Location of Violence and Sexual offences 2015-2017", fontname='Ubuntu', fontsize=18,
fontstyle='italic', fontweight='bold',color='green')
adjust_plot(ax)
analysis of crime based on Last outcome category
data['Last outcome category'].value_counts()
Investigation complete; no suspect identified 7568
Unable to prosecute suspect 1644
Status update unavailable 1613
Court result unavailable 783
Offender given a caution 591
Under investigation 531
Offender sent to prison 443
Offender given a drugs possession warning 249
Local resolution 245
Offender fined 198
Offender given suspended prison sentence 143
Offender given community sentence 138
Formal action is not in the public interest 129
Defendant found not guilty 116
Awaiting court outcome 104
Offender given penalty notice 85
Offender given conditional discharge 81
Court case unable to proceed 42
Offender otherwise dealt with 27
Suspect charged as part of another case 22
Further investigation is not in the public interest 21
Offender deprived of property 12
Action to be taken by another organisation 7
Offender ordered to pay compensation 5
Defendant sent to Crown Court 3
Offender given absolute discharge 1
Name: Last outcome category, dtype: int64
ax = data['Last outcome category'].value_counts().head(10).plot(kind='bar',figsize=(16,7),title='Crime Outome')
adjust_plot(ax)
turned out that most time suspect was unable to be indentified
Time series analysis of the crime
crime_time = data['Month']
crime_time.head()
0 2015-01-01
1 2015-01-01
2 2015-01-01
3 2015-01-01
4 2015-01-01
Name: Month, dtype: datetime64[ns]
crime_time = pd.DataFrame(crime_time)
#pd.to_datetime(crime_time['Month'],yearfirst=True)
#crime_time['Month'] = pd.to_datetime(crime_time['Month'],yearfirst=True)
crime_time.head()
| Month | |
|---|---|
| 0 | 2015-01-01 |
| 1 | 2015-01-01 |
| 2 | 2015-01-01 |
| 3 | 2015-01-01 |
| 4 | 2015-01-01 |
crime_time['value'] = 0
crime_time['Month'] = pd.to_datetime(crime_time['Month'],yearfirst=True)
crime_time.set_index(crime_time.Month,inplace=True)
del crime_time['Month']
crime_time.head()
| value | |
|---|---|
| Month | |
| 2015-01-01 | 0 |
| 2015-01-01 | 0 |
| 2015-01-01 | 0 |
| 2015-01-01 | 0 |
| 2015-01-01 | 0 |
crime_t = crime_time.resample('M')
crime_t
DatetimeIndexResampler [freq=<MonthEnd>, axis=0, closed=right, label=right, convention=start, base=0]
#
crime_per_month = crime_t.count()
ax = crime_per_month.plot(kind='line',figsize=(16,7),title='Monthly Times seties (2015-2017)')
adjust_pot(ax)
the above show crime was high around march of 2017
crime_w = crime_time.resample('Y')
crime_w = crime_w.count()
crime_w.index.name = 'Year'
crime_w.head()
| value | |
|---|---|
| Year | |
| 2015-12-31 | 6185 |
| 2016-12-31 | 6586 |
| 2017-12-31 | 5229 |
ax = crime_w.plot(figsize=(17,6),title='Yearly Time series of crime per year')
adjust_plot(ax)
there was more crime in 2016 than 2015 and 2017 but there was a particular month in 2017 that recorded highest crime which was around march
we also learnt that through out 2015 crime rate was on the increase
Mustapha Omotosho
constant learner,machine learning enthusiast,huge Barcelona fan