Insights from Predictive Policing - Tobias Brandt, Rotterdam School of Management

Explore the complexities of predictive policing in this comprehensive lecture by Tobias Brandt, Assistant Professor at Rotterdam School of Management. Delve into the intricate relationship between prediction quality and decision quality in smart city operations, focusing on data-driven decision-support for police forces. Examine how better prediction models don't always lead to improved outcomes, and understand the complex interplay between prediction models, resource allocation, and outcome measurement. Learn about an end-to-end framework for model selection that considers the entire process from data assessment to outcome evaluation. Cover topics such as crime explanation models, surveillance plots, 24-hour predictions, data sources, and the challenges of prescriptive analytics. Gain insights into the constraints and complex problems faced in implementing predictive policing strategies.

Introduction
What is predictive policing
Data sources
Crime explanation models
Surveillance plots
Predictions for 24 hours
What do I do with 24hour predictions
What do I do with crime predictions
How do we predict
Data overview
Summary
Average distance
Extreme case
Binary assessment
Distance analysis
Prescription by design
Complex problems
Constraints
Questions