Assessing Spatial and Temporal Aggregation of Point Level Data on Crime Hot Spot Forecasting

• In Event: Advances in Spatial Methods for the Study of Crime & Place

Abstract

This presentation will cover the theoretical (criminological and statistical) rationales for why aggregating point level data at different spatial and temporal resolutions can affect forecasting performance, metrics for assessing the effectiveness and efficiency of forecasts, and the results of varying spatial and temporal resolutions on the naïve model’s forecasting effectiveness and efficiency using 5.25 years of calls-for-data from the Portland Police Bureau. This research was conducted using operationally realistic spatial and temporal resolutions, along with the total area allowed to be labeled as a hot spot. The results will be discussed in terms of practical application along with a discussion of the (legal) importance of finding the proper balance in spatial and temporal resolutions.

Author

• Joel Hunt, National Institute of Justice