3. Local Policing Dashboard
Local Policing Dashboard
Must be relevant and current
Vulnerable Localities Index Recent crimes/incidents Demographics
Combine / cross
query disparate
data
Capture once – use
many times
4. Starting Point…..
• Requirement for web
map to support Local
Policing
• Jill Dando Institute VLI
methodology provided a
starting point
Review
Gap
Analysis
Solution /
Design
• Assess data
• Assess methodology
• Iterative Process
• Test > Review
• Finalise
6. Data selection
Income Deprivation
Burglary in a dwelling
Criminal Damage
Anti-social behaviour
Common Assault
Violence/ Wounding
Hate tagged incidentsEmployment Deprivation
15-24 year olds
Educational attainment
Truancy rates Mental Health
Child ProtectionAcorn Segmentation
Families first
Customer confidence
Gloucestershire Constabulary VLIJill Dando Institute VLI
Criminal Damage
7. data…..
• Demographics
Analysis: IMD vs commercial segmentation data
Deprived areas and high-
rise flats, 1757, 22%
Young families in low
cost private flats, 995,
12%
Struggling younger
people in mixed tenure,
658, 8%Social rented flats,
families and single
parents, 656, 8%
First time buyers in
small, modern homes,
555, 7%
Educated young people
in flats and tenements,
506, 6%
Elderly people in social
rented flats, 414, 5%
Singles and young
families, some receiving
benefits, 384, 5%
Low income large
families in social rented
semis, 288, 4%
Families in right-to-buy
estates, 286, 4%
10. Scoring system…
• Example of scoring: ASB
1 Year (rolling) ((ASB Rate per 1000 population)/(Community >0 Average))*100 Daily
3 Month Comparison
(LM = Last Month ASB Rate per 1000 population)
(M2 = 2 Months Ago ASB Rate per 1000 population)
(M3 = 3 Months Ago ASB Rate per 1000 population)
If LM 5%> M2 AND M2 5% > M3, Score = 25
If LM 5% > M2 AND M2 within or <5% M3, Score = 20
If LM within 5% of M2 and within or <5% of M3 Score = 10
If LM within or 5%< M2 and M2 5%> M3 15
If LM 5%<M2 and M2 within or < 5% M3 Score = 5
If LM 5%< M2 and M2 5%< M3 Score = 0
Monthly
Month to Date vs Same
Period Previous Year
M2D = Month to Date ASB Rate per 1000 population
PY = Same period M2D for previous year
If M2D 5%> PY Score = 20
If M2D with 5% PY Score = 10
If M2D 5%< PY Score = 0
Daily
11. Process flow…
•
GIS Database
Crime
Data
Incidents
Data
Demographic
Data
Criminal
Damage
Assault &
Battery
Violence /
Wounding
ASB
Harm
Hate
Selection Based onDate
Last Month
2 Months
Ago
3 Months
Ago
1 Year
(rolling)
Month to
Date
Month to
Date for
last Year
Count 1
Year
Count
Month to
Date
Count
Month to
Date Last
Year
Count Last
Month
Count Last
Month
Count Last
Month
Calculate
Score 3
Month Trend
Calculate
Score Trend
Month to
Date
Calculate
Score 1 Year
Postcode
VLI Table
Community
VLI Table
Postcode
Boundaries
Community
Boundaries
FamiliesFirst
Truancy
Child
Protection
Mental
Health
AddressBase
Premium
Dwellings
Calculate Score:
Demographics
Education
Calculate Score:
Families First
Count of
Child
Protection
Count of
Mental
Health
Calculate
Score Child
Protection
Calculate
Score
Mental
Health
Join
Attribute Join
Join
Attribute Join
Join
Spatial Join
Attribute Join
Selection Based onLocation
Iterate through
Communities or
Postcodes
Calculate Score:
Truancy
12. Process flow…
Daily ASB:
Community
1 Year
Rolling
Month to
Date
Compare
Temp ASB /
Community
Data
05:45 10:15
05:45
Script Start
Extract Relevant Records: Make Feature Layer
Iteration: arcpy.da.SearchCursor
Select per Community & Count
GetCount per Community & write to temp data
Community
VLI Table
05:45 10:15
06:40
Table Write
05:45 10:15
06:44
VLI Calc Monthly
Monthly VLI
Calculation
Daily VLI
Calculation
05:45 10:15
06:47
VLI Calc Daily
VLI Calc: Import numpy
Sum ASB: arcpy.da.TableToNumPyArray
Count Communities ASB >0: Select by Attribute, GetCount
Join: Temp table to VLI
Calculate 1 Year:
((float(!VLI_ASB_Y1.Temp.COUNT!))* (1000/!Community.Population!)) / (" + str(sumASB) + "/" + str(YRASB_Count) + ")* 100"
Calculate Month to Date Comparison:
CalculateField: (Table, "Community_VLI.M2D_ASB",
"qtrASB(float(!VLI_ASB_PY.TEMP.COUNT!* (1000/ !Community.Population! )), float(!VLI_ASB_M2D.TEMP.COUNT!* (1000 / !Community.Population! )))",
"PYTHON_9.3",
"""def qtrASB(PY, M2D): n n
if (M2D > ((PY/ 100) * 105)): n
return 20 n
elif (M2D < ((PY/ 100) * 105) and M2D > ((PY/ 100) * 95)): n
return 10 n
else: n
return 0 n""")
Calculate VLI Total: Sum all categories
05:45 10:15
Daily ASB:
Postcode
1 Year
Rolling
Month to
Date
Compare
Temp ASB /
Postcode
Data
07:30
Start: Postcode
Extract Relevant Records: Make Feature Layer
Tabulate Intersection & GetCount
05:45 10:15
Daily:
Repeats
08:50
Repeats
05:45 10:15
Daily VLI
Calculation
Postcode VLI
Table
Monthly
ASB:
Community
3 Month Trend
Temp ASB /
Community
Data
05:45 10:15
06:05
Script Start (Monthly)
05:45 10:15
Monthly
ASB:
Postcode
3 Month Trend
Temp ASB /
Postcode
Data
09:00
Start: Postcode Monthly
05:45 10:15
Monthly VLI
Calculation
10:05
Calculate Postcode Monthly
As a police force, keeping our communities safe from harm is one of our main objectives.
The Dashboard aims to make best use of data in order to support Local Policing
Understanding vulnerability within our communities is key, and GIS is an effective way to get to grips with the layers of complexity
Provide a platform to support local policing:
Relevance Making sure that only relevant data are included – don’t muddy the waters…
Currency - It’s key to ensure the currency of the data if the Dashboard is going to be effective in supporting decision making, and to ensure currency, automating data processing is essential to ensure currency but consistency.
Identifying vulnerable communities & social incohesion:
Calculate a Vulnerable Localities Index (VLI)
Recent crimes / incidents (specific types)
Demographics & behavioural
Make best use of available data:
Capture once, use many times…
Combine / cross query disparate data
The methodology set out by the Jill Dando Institute provided a sound basis for the work, but it’s important to question the methodology in order to; improve and update, make relevant for the organisation
A review of the recommendations was made together with gap analysis, with a resolution & design
Design, set up steering group, revisit/scrutinise design/ideas, demo – iterative process
Automate: Python and scheduled tasks
JDI uses IMD to LSOA. Need to assess which segmentation groups best fit in with sentiments of the JDI study.
Step 1. Look for correlations geographically
2. Identify segmentation types for use in the VLI,
3. Apply the types across the county, - greater granularity, increased currency
Talk through interface
Talk through interface
Additional scoring to compare last 3 months and same period for last year,
Same process applied to other crime and incident data
Scoring applied to the other data sets; truancy, families first, etc