Analysis and Prediction of Crime Using Machine Learning Techniques

Abstract

Crime is an exceedingly alarming factor in India and it should be regulated at all costs. We are now able to use various strategies to reduce the crime rate in our nation, India, with the development of technology. This paper focuses on the detection of crime using deep learning and machine learning techniques. Pre-processing technique has been used to convert the data into a suitable format for analysis and for detection. Primarily focus on providing a forecast of the type of crime that may happen built on the place where it has now shown location. Machine learning has been used to build a model with the use of a training dataset that has gone through the data maintenance and renovation process. The proposed methodology is to analyze detection by making the use of machine learning processes such as decision tree and support vector machine for detection and also explored the k-nearest neighbors and artificial neural networks techniques to predict the crime, and the relevant authorities should then take precautionary steps to remain alert and monitor the situation. The characteristics of data are analyzed and realized by using data visualization tools. The algorithms are implemented in python language, and after training, the model performed testing using test data, and results were pretty good. The precision rate is achieved at 92% using neural network as compared with other algorithms, and also the processing time for confirmation is lowest with the neural network algorithm. Overall, of the three selected algorithms, the linear regression algorithm performed the best. The purpose of the work is to determine the precise and effective machine learning algorithms used in data analysis and in the prediction of models of violent crime.