Prophet for crime=machine learning + human behavioural data
The normal doctor cures; The good doctor predict; The best doctor prevents.
Crime is a well-known social problem affecting the quality of life and the economic development of a society. As a chronic disease of a society, it is not enough to cope with it when it shows up. The exist unreported crimes and potential crimes, like a lurking illness, is what really affects the future of society.
Using machine learning models with anonymized and aggregated human behavioural data (compute real-time demographic using phone calls data), we could achieve as high as 70% accuracy of prediction about where the crime will happen in the next month/ next week.
Predictive Policing enables police officers to put their effort into areas that crimes are more likely to happen as well as deter the potential crimes. Even though the newly predicted areas may not have seen recent crimes, if they are similar enough to prior ones, they could be considered to be high-risk areas to monitor closely. This is an important advantage given that in some areas people are less inclined to report crimes. Moreover, it can identify novel risk-inducing or risk-reducing features of geographical areas (e.g. the presence of liquor stores, certain types of major stores, bars, etc.) hence improve the efficiency of policies to identify the potential crime area.
Reference : https://arxiv.org/pdf/1409.2983.pdf
