SLIDE 1
Al Algorit ithms & Explanatio ion: : A A Humble Framin ing - - PowerPoint PPT Presentation
Al Algorit ithms & Explanatio ion: : A A Humble Framin ing - - PowerPoint PPT Presentation
Al Algorit ithms & Explanatio ion: : A A Humble Framin ing Jeremy Heffner HunchLab Product Manager & Senior Data Scientist jheffner@azavea.com Predictive Policing Prevent Crime Design patrol allocations to minimize the predicted
SLIDE 2
SLIDE 3
Prevent Crime
SLIDE 4
Design patrol allocations to minimize the predicted preventable reported harm
SLIDE 5
SLIDE 6
SLIDE 7
SLIDE 8
SLIDE 9
SLIDE 10
SLIDE 11
“The more powerful you are, the more your actions will have an impact on people, the more responsible you are to act humbly.” Pope Francis
SLIDE 12
Models make mistakes
SLIDE 13
SLIDE 14
Time Since Last & Theft From Vehicles (Seattle)
SLIDE 15
Wind Speed & Aggravated Assault (Chicago)
SLIDE 16
MVT and Distance from School (Philadelphia)
SLIDE 17
Trade Secrets
SLIDE 18
Is your creation that special that protecting it trumps the public interest? Likely no.
SLIDE 19
Decision / Allocation Policy Explanations
SLIDE 20
SLIDE 21
# Assaults x 12 x 40% # Burglary x 8 x 60% # MVT x 5 x 65% # Larceny x 3 x 50% # Robbery x 10 x 40% Sum to Predicted Preventable Harm
SLIDE 22
SLIDE 23
Dealing with Uncertainty / Randomness
SLIDE 24
101 100 2 2 2 50 1 1 1
SLIDE 25
101 100 2 2 2 50 1 1 1
SLIDE 26
75 30 2 2 2 60 1 1 1 101 100 2 2 2 50 1 1 1 80 60 2 2 2 40 1 1 1
SLIDE 27
75 30 2 2 2 60 1 1 1 101 100 2 2 2 50 1 1 1 80 60 2 2 2 40 1 1 1
SLIDE 28
This location was predicted to experience 0.01 robberies, 0.02 burglaries, … which represents 40 units of predicted preventable harm. This level of harm is 2 standard deviations above the mean, resulting in 4 lottery entries for this location. 5 locations were desired for patrol out of 11 eligible locations in the beat. Based upon the analysis, this location would be selected 83% of the time under the same conditions.
SLIDE 29
10
SLIDE 30
10 7
SLIDE 31
10 7
SLIDE 32
10 7
SLIDE 33
10 7
SLIDE 34
10 7
Acknowledges uncertainty Introduces randomness in explanation Preserves diversity in training examples
SLIDE 35
Humans make mistakes
SLIDE 36
SLIDE 37