ETD

• contextual data
• crime prediction
• forecasting
• machine learning
• predictive policing

Modern data mining approaches can be used to analyze data regarding past criminal events, opening up chances for prediction algorithms in this field. Unfortunately, lack of context information on crimes reasonably limits the efficacy of prediction models. In this work, we investigate what type of context information should be gathered and exploited in real-world settings, and how it can be analyzed along with ordinary data, to implement novel AI-based software solutions for more effective crime prediction tools.