Journal of Computer, Internet and Network Security ISSN: 2457-0176 (Online)

Authors: Jay Bhanushali, Karan Bhanushali, Darshil Hariya, Theres Bemila Jenet

Abstract: The increasing rate of criminal activities is a growing concern for any particular country/region. The proposed system intends to develop a user-friendly web application for the stake-holders such as Invigilators, NGOs, and end-users. The system establishes a simple relation between the above-mentioned stake-holders where any individual user can report a crime without himself going to the police station. The invigilators then can track the complaint and give an option to the NGO for Rehab. Also, this system analyse crime data of India scrapped through various websites. The main focus is to predict the crime which is most likely to occur in the near future using various Machine Learning models. The parameters which will be required for the proposed model are timestamp which includes hours, days, months, years, geographical location, i.e., latitude & longitude and the type of crime that occurred. The proposed system implements various historical Machine Learning algorithms like KNN (K- nearest neighbours), Decision trees and Random forest. The goal is to analyse the surges and the hotspots of the crime so that it is efficient for officials to enhance security in such locations. To get a clear understanding of the patterns of criminal behaviour, this will help to carry the investigation sooner. The existing system predicts the different crime categories in different locations using similar historical Machine Learning models like Naïve Bayes, Decision Tree, etc. The proposed system attempts to make the classification work more sensible by merging similar classes into larger classes. Finally, the proposed system also provides an opening for future work like a prediction of the crime rate in the near future.