From Prototype to Production: Model Implementation

• In Event: Advances in Machine Learning and Crime

Abstract

Traditional criminal justice data analysis takes place on static data sets, which precludes implementation or validation of findings on ‘live’ data. As the collection and use of data within criminal justice agencies expands, prediction models are shifting from academic exercises to integral pieces of agency operations. The infrastructure behind one large agency’s data systems, and implementation of a machine learning algorithm are discussed.

Author

• Connor Concannon, John Jay College of Criminal Justice / CUNY Graduate Center