Aligning Prediction and Decision Quality in Smart City Operations: Insights from Predictive Policing

This is a talk by Tobias Brandt, Assistant Professor of Business Information Management at Rotterdam School of Management, Erasmus University.

In this project, we analyze the relationship between prediction quality and decision quality, which reflects the greater phenomenon of prescriptive analytics increasingly replacing predictive approaches. Using the case of data-driven decision-support for police forces, we demonstrate that a model with a better prediction quality does not automatically lead to better eventual outcomes if decisions are based on this model. We illustrate that this effect is caused by the complex relationship between the prediction model, the allocation of the constrained resources (patrol cars) based on these predictions, and the quality measurement of the eventual outcome, and varies especially with the latter. To alleviate these issues, we propose an end-to-end framework for model selection that takes into account the full process from assessing available data to the outcome measurement.

10:45 - 11:00 - Registration

11:00 - 11:45 - Aligning prediction and decision quality in smart city operations - Tobias Brandt

11:45 - 12:00 - Q&A

Registration is from 10:45, talk starts at 11:00.

The Alan Turing Institute is located on the first floor (right hand side) of the British Library and can be accessed via the main doors from 09:30 onward. Please allow 15 minutes for queuing when planning your journey.