Analytical Study of Some Selected Classification Algorithms and Crime Prediction

To prevent the crime these days police exercises particularly in the case of investigation, emphasis on Artificial Intelligence, data mining and Machine learning aspect. To prevent future crimes it is necessary to understand the crime behavior from the earlier crime records. The more numbers of recorded crime data approaches the research to analysis the data for prediction and prevention of crime. Most of the researcher use to cluster the data for further classification to predict the definitive of crime. To prevent the crime these tools are xeric applicable to predict the most sensitize zone in the city. Thesis concentrated on the methods to predict the crime and on his hidden arrays in the existing past records. The objectives of the thesis is to predict the certain possibility of crime by applying data ruining approach through WEKA are applied to confirm criminality y proclamation. three algorithms are referred from different groups of methods: SMO Zero R and J 48 decision trees. Over 10000 records from Indian police department are collected to predict the frequency of crime in overall and its behavior in which Naive algorithm shows the reliable prediction represent against crime frequency. This paper compares the three diverse order classification to be specific, SMO, Zero R and J 48 Decision Tree for anticipating Crime Category' for various states in India. The outcomes from the examination demonstrated that, Decision Tree calculation out performed calculation and sieved 41.44% and 73.33%% Accuracy in anticipating Crime Category for various conditions of India. order classification to be specific, SMO, Zero R and J 48 Decision Tree for anticipating Crime Category' for various states in India. The outcomes from the examination demonstrated that, Decision Tree calculation out performed calculation and sieved 41.44% and 73.33%% Accuracy in anticipating Crime Category for various conditions of India.