Analyzing the Spatial Distribution of Property Crime in the Annapolis County Brad Benham Major Research Project in GIS for Business Project Sponsor: Annapolis County RCMP
Jan 27, 2015
Analyzing the Spatial Distribution of Property Crime
in the Annapolis County
Brad Benham
Major Research Project in GIS for BusinessProject Sponsor: Annapolis County RCMP
Outline• Previous Projects Analyzing Crime• Project Purpose• Data Processing• Mapping Crime by Community• Statistical Analysis• Socio-economic & Socio-demographic
Variables• Predicting Crime• Hot Spot Analysis• Conclusions• Limitations and Recommendations• Return on Investment
Previous Projects Analyzing Crime
• GIS Implementation for Crime Analysis and Community Policing by Melanie Foote (1999)(Royal Canadian Mounted Police)
Project Purpose• GIS has never been used to analyze crime in the Annapolis Valley• The information GIS could extract from crime data could hold
significant value to police in the Annapolis County which is why the RCMP has sponsored this project by providing property crime data from 2013
• The purpose of this project is to create a database of all property crime events in the Annapolis County and to perform several queries and geographical analysis to extract information from the data
• The result will help in the understanding of who, what, where, when, and why crime occurs
• Furthermore, relationships between types of crimes, their location, time, and some socio-economic/socio-demographic characteristics of the region will be explored and used to help predict crime in the future
Defining Property Crime• Property crime is defined in many different ways and the types of
crime that are categorized by property crime are highly variable depending on the source. Property crime generally includes taking money or property when there is no force of threat of force against the victims (National Institute of Justice, 2013).
• Breaking and entering is defined as entering a residence or other enclosed property through the slightest amount of force, without authorization.
• Theft is a generic term for all crimes in which a person intentionally and fraudulently takes personal property of another without permission or consent and with the intent to convert it to the taker’s use (including potential sale).
• Mischief is a specific injury or damage caused by another person’s action or inaction. When mischief is malicious, it is a criminal act involving reckless or intentional behaviour such as vandalism.
Explaining Crime
• Routine Activities Theory– Crime occurs when there is an intersection in time and
space of a motivated offender, an attractive target, and a lack of capable guardianship. People’s daily routine activities affect the likelihood they will be an attractive target who encounters an offender in a situation where no effective guardianship is present. Changes in routine activities in society (e.g., women working) can affect crime rates (ibid).
Input Data
Field Name Definition Example[ID] Unique Identifying Number 1, 2, 3...999[Unit] Responding Police Unit Middleton, Annapolis[Crime] Type of Crime Break and Enter[Type] Detailed Crime Type B&E into Residence, Cottage [Year] Year the Crime Occurred 2013[Month] Month the Crime Occurred January, February, etc.[Day] Day the Crime Occurred Monday, Tuesday, etc.[Time] Time the Crime Occurred 16:45, 20:00, etc.[Address] Street the Crime Occurred on Main St., Paradise Lane[Community] Town the Crime Occurred in Bridgetown, Lawrencetown[Postal_Code]* Postal Code for the Location B0S1M0[County] County where Crime Occurred Annapolis County
Example of Raw Data:3) H DIV BRIDGETOWN OFFICE Break and Enter - Residence 348(1) CC (FIP) 2013/01/02 20:30 MAIN STREET LAWRENCETOWN, ANNAPOLIS COUNTY NS Canada4) H DIV MIDDLETON OFFICE Break and Enter - Other 348(1) CC (FIP) 2013/08/10 09:26 10 HIGHWAY, NICTAUX, ANNAPOLIS COUNTY NS Canada5) H DIV ANNAPOLIS CO STREET CRIME ENFORCEMENT UNIT Break and Enter - Residence 348(1) CC (FIP) 2013/08/12 12:31 MOUNT HANLEY ROAD, MOUNT HANLEY, NS Canada
• Missing House Numbers
Using Network Analyst
• Road Network with 10km buffer– Calculate geometry of road segments (Length)– Assign speed limits (Speed in km/h)
• Local (50)• Arterial & Collector(80)• Expressway/Highway (100)• Ramp (40)
– SECONDS = [Length] * 3.6 / [Speed]
Closest Facility Solver
Driving Time and Distances to Road Junctions from RCMP Offices
Geocoding Crime Incidents
Closest Facility to Crime Incidents
Driving Time to Crime Incidents
Crime Rate per 100 People by Dissemination Areas
Mapping Crimes with Proportional Symbols by Address
Mapping Crime by Community
• Median Center tool used to find communities
• Frequency tool used to find crime counts
Mapping Crime by Community
Mapping Crime by Community and by Crime Type
Communities with the Highest Number of Crimes
Concentration of Various Crimes
Adding Information Classes
• Month Name• Season– Spring, Summer, Autumn, Winter
• Time of Day– Morning, Afternoon, Evening, Night
• Weekend/Weekday
Mapping Crime in Communities by Season
Summary of Seasonal Crime
Spring Summer Autumn Winter
Break and Enter 52 58 29 42
Mischief 70 96 69 30
Theft 97 71 115 46
Total 219 225 213 118
25
75
125
175
225
52 58
2942
7096
69
30
9771
115
46
219 225213
118
Frequency of Crime Incidents by Type and Season
Break and Enter
Mischief
Theft
Total
Season
Num
ber o
f Crim
es
Mapping Crime in Communities by Time of Day
Summary of Crime by Time of Day
Morning (6AM -
12PM)
Afternoon (12PM - 5PM)
Evening (5PM -
10PM)
Night (10PM -
6AM)
Break and Enter
49 59 41 32
Mischief 60 75 63 67
Theft 116 129 60 24
Total 225 263 164 123
25
75
125
175
225
275
49 5941 32
6075 63 67
116 129
6024
225
263
164
123
Frequency of Crime Incidents by Time of Day
Break and Enter
Mischief
Theft
Total
Time of Day
Num
ber o
f Crim
es
Mapping Crime in Communities by Weekdays/Weekends
Summary of Crime on Weekdays/Weekends
Analyzing Association
• Associations (higher or lower counts than expected)– Crime is associated with seasons, time of day, weekends,
and office– Office is associated with seasons– Seasons are associated with time of day– Time of day is associated with weekends
• No Association (even distribution)– Office is not associated with time of day or weekends– Season is not associated with weekends
Analyzing CorrelationsCorrelations Value SignificancePercent of households headed by a lone parent
0.462**
Very Significant (0.01)
Population density per dissemination area
0.433**
Very Significant (0.01)
Males aged 15-24 as percentage of total population
0.375* Significant (0.05)
Average $ spent on games of chance per person
0.374* Significant (0.05)
Percent of population 15+ that is divorced
0.371* Significant (0.05)
2013 Average $ spent on Alcohol per person
0.352* Significant (0.05)
2013 Average household expenditure (curr $)
0.349* Significant (0.05)
2013 Average household income (constant $)
0.333* Significant (0.05)
Percent of shelter paid by rent 0.325* Significant (0.05)Percent of household population with only high school certificate
0.292* Significant (0.05)
Choosing the Best Model to Predict Crime
Map of Residuals
Is Crime Clustered or Dispersed?
Crime Type Observed Mean Distance (m)
Expected MeanDistance (m)
Z-score (Std. Dev.)
Ratio
Break and Enter 823 2168 -15.95 0.37Theft 334 1604 -27.45 0.2Mischief 424 1707 -23.4 0.24All Crime 148 1061 -45.84 0.13
Hot Spot Analysis
Conclusions (1)• Lawrencetown experienced the most break and enters
(increase during Christmas holidays)• Break and enters occur more frequently on the North
Mountain and the South Mountain than other crimes• Winter season experiences only 15% of the years crime• Crime is most often committed at night and during the
morning• Crime is more commonly commit on the weekend. Especially
mischief but theft much more often on weekdays• The workload among Bridgetown and Middleton offices is
distributed evenly
Conclusions (2)• A high percentage of population that is young male and a high
divorce rate were found to be positively correlated with crime which is consistent with academic literature .
• The two variables most positively correlated with crime were percent of lone parent households and population density
• The two variables found to be most important when predicting crime were average household income and population density
Limitations and Recommendations
• Geocoding• Missing house numbers• Confidentiality privilege could be given
• Centroid of postal codes• Communities• Median center tool
• Crime report times• Annapolis Royal police station
Return on Investment
• In Canada, the value of property stolen or damaged due to property crime in 2004 was over $5.7 billion dollars.– A reduction in property crime can help save a significant
amount of citizens money– Tax dollars can be saved by allocating police more
efficiently to increase conviction rate and improve safety.
Other Benefits of GIS
• Less damage and loss of property• Increase in citizen safety• Increase attractiveness to live in Annapolis
County• Increase public awareness and engagement in
crime prevention (Neighbourhood Watch programs)• Predict crime changes over time• Re-evaluate police tactics and strategies to
improve services
Thanks for Listening
• Questions?