PREDICTIVE POLICING: A Machine Learning Approach to Predict and Control Crimes in Metropolitan Cities

Security is the one of the basic need of human’s life and biggest challenge of history that cannot be diminished at least in metropolitan cities like Karachi. It can only be controlled by efficient resources allocation and effective strategies with forthcoming insight of criminal moves. Big data analytics with the support of machine learning algorithms makes it possible to deals with huge amount of data, extract hidden inter-connection, pattern and meaningful information. This paper, proposed the model for the predictive policing system and built test model using k-means and naïve Bayes methodologies for street crime in Karachi region. The model is then run under R and WEKA environment which produced accuracy around 70%.