Crime prediction and forecasting in Tamilnadu using clustering approaches

Crime is one of the most predominant and alarming aspects in our society and its prevention is a vital task. Crime analysis is a systematic way of detecting and investigating patterns and trends in crime. In this work, we use various clustering approaches of data mining to analyse the crime data of Tamilnadu. The crime data is extracted from National Crime Records Bureau (NCRB) of India. It consists of crime information about six cities namely Chennai, Coimbatore, Salem, Madurai, Thirunelvelli and Thiruchirapalli from the year 2000-2014 with 1760 instances and 9 attributes to represent the instances. K-Means clustering, Agglomerative clustering and Density Based Spatial Clustering with Noise (DBSCAN) algorithms are used to cluster crime activities based on some predefined cases and the results of these clustering are compared to find the best suitable clustering algorithm for crime detection. The result of K-Means clustering algorithm is visualized using Google Map for interactive and easy understanding. The K-Nearest Neighbor (KNN) classification is used for crime prediction. The performance of each clustering algorithms are evaluated using the metrics such as precision, recall and F-measure, and the results are compared. This work helps the law enforcement agencies to predict and detect crimes in Tamilnadu with improved accuracy and thus reduces the crime rate.