Crime Detection using Data Mining

Abstract

As crime rates keep spiralling each day, new challenges are faced by law enforcement agencies. They have to keep their forces on the lookout for any signs of criminal activity. This may only cause more burden on their resources. The law enforcement agencies should therefore be able to predict such increases or decreases or trends in crime, such as the approximate number of murders, rapes, thefts, or any such crimes that may occur in a particular area in a particular month, year, or any timespan, or, the overall number of crimes occurring in a country in a particular year in the future, or any other prediction or projection of future crime statistics. First, our system proposes to extract data from crime record repositories, on which we intend to perform data mining. Data classification and regression algorithms then help in forecasting and predicting this is proposed to be done by first training a set and then applying the learned rules on the test set in order to determine the predicted output. Using this, law enforcement agencies can better understand how the crime pattern across a certain region, or interval of time is, and using this data, such agencies can take proactive action to stem the rise of particular crimes in particular regions, or during particular times. This would save them a lot of time, money and effort. Our system proposes to mine this data and thus run appropriate algorithms on such data. This predicted output could also be presented to the user in the form of clusters using a data visualization algorithm like K-means clustering algorithm. The final end-product could thus be a system where some future predictions would made by training crime data sets, and the output could be visualized in order to be simple to comprehend for the user.