The Los Angeles Predictive Policing Experiment

Charlie Beck, Chief of Police
Los Angeles Police Department
2
3
4
Predictive Policing Time Horizon
Frequency of Statistical Review

1995

Monthly

1990
Annually

5

2000

Near
Real Time

2...
Predictive Policing, Our Definition
•
•
•
•
•
•
•

6

“A place-based approach to crime analysis that utilizes
algorithm-dr...
How is this different?
•
•
•
•
•
•
•
•
•
•

7

Evidence-based rather than heuristic
Forecasting versus Retrospective Comps...
The Los Angeles Predictive Policing Experiment
Questions to be answered…

8
Questions to be answered…
Is it possible to predict crime?
If we predict it, can we prevent it?
If we prevent it,
how can ...
Jeff Brantingham
UCLA Anthropology
George Mohler
Santa Clara Mathematics
George Tita
UCI Criminology, Law and Society
10
Repeat Victimization
probability p(τ)

0.006

repeat crime is much more likely to
happen in a short interval of time after...
Research Design: The LAPD Experiment
• Rigorous examination of both forecasting value
and of dosage
• Careful design of ex...
Daily Randomization
• Foothill Division serves as both the treatment and
control area (seamless to the subjects)
• On rand...
14
Foothill Patrol Area

15
Forecasting Tool Interface v 1.0

16
Daily Forecast Map

17
Close-up View of Mission “Boxes”

Pierce St and Borden Ave
RD1613
18

Carl St between Borden & Chivers Ave
RD 1613
The Timeline
• October 19, 2011:
Training on production of forecasts and methodology begins

• November 6, 2011:
Three mon...
Initial Observations…
•
•
•
•
•
•
•
•
•
•
21

Bottomline:
How accurate are the forecasts?
Versus random and versus state o...
Preliminary results
• soft numbers pending completion of statistical study
• focusing on weekdays only
• accuracy of PredP...
23

Predictive Policing is promising in historical data analysis
Crime Numbers (Burg, BFMV & GTA)
Test Period Weeks 1 thru 5 - 2010 v. 2011

80

70

66

67

65
60

2010 avg.= 60

60

51

...
Securing Buy-In from Patrol

Get Creative…
25
Patrol Activity (addressing forecasted mission areas)
Test Period Weeks 1 thru 5

140

= Hours “in the Box” per Week

120
...
New Version of Software…
•
•
•
•

28

Weekend vs. weekday finding
Ease of use improvements
4 times per day by watch
Roll o...
29
30
31
The Los Angeles Predictive Policing Experiment

Charlie Beck, Chief of Police
Los Angeles Police Department
Additional Slides if Needed

33
Foothill “dosage” evidence:

Number of Unique Boxes Patrolled

1SD
mean
1SD

34
1SD
mean
1SD

35
1SD

mean
1SD

36
Santa Cruz
• Very basic crime analysis prior to Predictive Policing
• Updated daily, the prediction maps identify “boxes” ...
42
43
PredictivePolicing
Upcoming SlideShare
Loading in …5
×

PredictivePolicing

1,158 views

Published on

Politiechef Charli Beck uit LA

Published in: Technology
0 Comments
5 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,158
On SlideShare
0
From Embeds
0
Number of Embeds
46
Actions
Shares
0
Downloads
56
Comments
0
Likes
5
Embeds 0
No embeds

No notes for slide

PredictivePolicing

  1. 1. The Los Angeles Predictive Policing Experiment Charlie Beck, Chief of Police Los Angeles Police Department
  2. 2. 2
  3. 3. 3
  4. 4. 4
  5. 5. Predictive Policing Time Horizon Frequency of Statistical Review 1995 Monthly 1990 Annually 5 2000 Near Real Time 2005 Real Time 2010 Forecasting
  6. 6. Predictive Policing, Our Definition • • • • • • • 6 “A place-based approach to crime analysis that utilizes algorithm-driven crime forecasts to inform decision making to prevent crime” Not individual-based Not arrest-based Risk-based Deployment of patrol resources Builds on community policing Builds on Hot Spots Offers more specificity Sets us up for better long term strategic analysis
  7. 7. How is this different? • • • • • • • • • • 7 Evidence-based rather than heuristic Forecasting versus Retrospective Compstat Look Reduces tendency to “chase the dots” Offers more specificity in time and space Should eliminate biases (arrests are not used) Leverages existing real-time data Not a massive multivariate model Uses Date, Time, Place and Type of crime Allows cops to problem solve in right place at right time Dosage findings should increase efficiency
  8. 8. The Los Angeles Predictive Policing Experiment Questions to be answered… 8
  9. 9. Questions to be answered… Is it possible to predict crime? If we predict it, can we prevent it? If we prevent it, how can we measure that? 9
  10. 10. Jeff Brantingham UCLA Anthropology George Mohler Santa Clara Mathematics George Tita UCI Criminology, Law and Society 10
  11. 11. Repeat Victimization probability p(τ) 0.006 repeat crime is much more likely to happen in a short interval of time after the first event 0.005 0.004 0.003 0.002 0.001 0 0 50 100 150 200 250 300 350 days between repeat crimes (τ) 11 400
  12. 12. Research Design: The LAPD Experiment • Rigorous examination of both forecasting value and of dosage • Careful design of experiment is crucial to determine if Predictive Policing “works” • Several designs possible, but only one – randomization – represents the “gold standard” of scientific design 12
  13. 13. Daily Randomization • Foothill Division serves as both the treatment and control area (seamless to the subjects) • On random treatment days, the “daily mission” is determined by the Predictive Policing model; otherwise, mission is determined using existing methodology • Experiment will run for a minimum of 6 months and evaluate whether total crimes lower on days when Predictive Policing model used 13
  14. 14. 14
  15. 15. Foothill Patrol Area 15
  16. 16. Forecasting Tool Interface v 1.0 16
  17. 17. Daily Forecast Map 17
  18. 18. Close-up View of Mission “Boxes” Pierce St and Borden Ave RD1613 18 Carl St between Borden & Chivers Ave RD 1613
  19. 19. The Timeline • October 19, 2011: Training on production of forecasts and methodology begins • November 6, 2011: Three month randomized study begins • March 2012: Evaluation results Continued iterative process to test both the algorithm as well as the dosage and interventions. 20
  20. 20. Initial Observations… • • • • • • • • • • 21 Bottomline: How accurate are the forecasts? Versus random and versus state of the art How effective in crime fighting Crime numbers time series Dosage discussion Weekend vs. weekday finding Ease of use improvements 4 times per day by watch Roll out plans
  21. 21. Preliminary results • soft numbers pending completion of statistical study • focusing on weekdays only • accuracy of PredPol algorithm vs. best-practice • PredPol vs. independent control group: 10.6% vs. 9.8% • PredPol applied to control group: 11.4% vs. 9.8% • PredPol is 8-16% more accurate compared to best practice 22
  22. 22. 23 Predictive Policing is promising in historical data analysis
  23. 23. Crime Numbers (Burg, BFMV & GTA) Test Period Weeks 1 thru 5 - 2010 v. 2011 80 70 66 67 65 60 2010 avg.= 60 60 51 2011 avg.= 50 50 43 40 37 40 40 42 10 2011 20 2010 30 0 week 1 24 week 2 Test Period 2010 avg.= 60 week 3 week 4 week 5 Test Period 2011 avg.= 42
  24. 24. Securing Buy-In from Patrol Get Creative… 25
  25. 25. Patrol Activity (addressing forecasted mission areas) Test Period Weeks 1 thru 5 140 = Hours “in the Box” per Week 120 120 = Checks per Week 96 100 72 80 Occupy LA 51 60 47 40 42 40 27 20 18 18 0 1 26 2 3 Avg Hours in the Box per Week = 35 hrs 4 5 Avg Checks per Week = 71 checks
  26. 26. New Version of Software… • • • • 28 Weekend vs. weekday finding Ease of use improvements 4 times per day by watch Roll out plans
  27. 27. 29
  28. 28. 30
  29. 29. 31
  30. 30. The Los Angeles Predictive Policing Experiment Charlie Beck, Chief of Police Los Angeles Police Department
  31. 31. Additional Slides if Needed 33
  32. 32. Foothill “dosage” evidence: Number of Unique Boxes Patrolled 1SD mean 1SD 34
  33. 33. 1SD mean 1SD 35
  34. 34. 1SD mean 1SD 36
  35. 35. Santa Cruz • Very basic crime analysis prior to Predictive Policing • Updated daily, the prediction maps identify “boxes” to be patrolled “when free” • Promising results, but limited ability to do careful evaluation • is this because Santa Cruz is now doing crime analysis where before they were not? • no controlled comparisons 41
  36. 36. 42
  37. 37. 43

×