Crime Rate Prediction with Region Risk and Movement Patterns Shakila Khan Rumi RMIT University Melbourne, VIC shakilakhan.rumi@rmit.edu.au Phillip Luong RMIT University Melbourne, VIC phillip.luong@rmit.edu.au Flora D. Salim RMIT University Melbourne, VIC flora.salim@rmit.edu.au Abstract—The location-based social network, FourSquare, helps us to understand a city’s mass human mobility. It provides data that characterises the volume of movements across regions and Places of Interests (POIs) to explore the crime dynamics of a city. To fully exploit human movement into crime analysis, we propose the region risk factor which combines monthly aggregated crime and human movement of a region across different time intervals. We then derive a number of features using the region risk factor and conduct extensive experiments with real world data in multiple cities that verify the effectiveness of these features. I. I NTRODUCTION One of the basic demands for every person in society is a safe and secure living space. Therefore, it is important to find ways to control a city’s crime rate, which hinders economic development. Understanding the root causes of what increases the likelihood of crime occurring at any time has great benefits for law enforcement to coordinate response and prevention strategies. According to criminology theory, the surrounding environment, including neighbourhood regions, and the movement of people play a crucial role in crime event prediction. The widespread use of location-based social networks such as FourSquare open up a door of opportunities to analyse crime events in a timely manner. In this paper, we study crime rate prediction with the help of urban mobility data. Recently, there has been research linking crime events with urban dynamics using FourSquare data [1]. However, this study focused on a region’s check-in information to predict crime events. There has been no focus on linking human movement between two regions. In [2], the authors considered the movement between regions using taxi flow data for crime inference, however, did not account for the variation of move- ment in different time periods of day. For example, the people who move from their home to work will move in the opposite direction in the afternoon. In this paper, we analyse crime inference with human mobility at different periods of the day. In Figure 1, we observe the difference in the number of people moving in the morning and afternoon for New York City. The changes of human movements between the two periods of the day are highlighted with different colour. To fully exploit the human dynamics in crime inference, we further introduce Region Risk, which associates crime with people moving in that region over a time interval. The hypothesis is that human mobility from a high crime risk area implies a high crime risk in the arrival area. We derive numerous features with the assistance of Region Risk. The significance of this and other features are verified for Chicago and New York City at different times of the day. The contributions of this paper are summarized as follows: • This is the first work that predicts the crime rate based on the dynamic features that associate Region Risk and movement patterns between regions. • New features associated with Region Risk and human mobility are crafted. • This work verifies the effectiveness of different features in the crime inference problem using correlation and re- gression analysis. Real-world crime data and FourSquare movement data are used for evaluation. The experimental results show that the Region Risk features are highly correlated with crime count of a region. 2859 26 3817 10 90 3338 13 3339 3818 29 4297 12 24 7167 19 17 15 3340 135 23 21 73 24 10 34 16 176 13 12256 5252 64 11 10 10 20 10 (a) Morning 3817 147 3338 11 3339 3818 4297 20 7167 19 17 16 3340 69 178 24 27 46 567 19 16197 5252 110 14 21 27 52 38 10 15 10 13 49 11 40 2859 18 12 (b) Afternoon Fig. 1: Check-in movements for a few venues in New York in the mornings and afternoons of March, 2018 II. RELATED WORK Previous data mining studies were conducted to verify the impact of human mobility onto crime. In [3], the authors extracted human behavior from mobile network activities and demographic features of people connected to the network over different regions and times. The study showed that the combination of mobile activity data and demographic data can be used to predict crime events with better accuracy. FourSquare check-in data measures the ambient population of a region and is used to understand the long-term crime arXiv:1908.02570v1 [cs.SI] 25 Jul 2019