| Popis: |
In the era of globalization, crime is a subject that is feared and dreaded by the public all around the world. Crime is an integral part of risks that we are facing it in every day. Nowadays, there have been many criminal cases reported by the mass media, such as theft, rape and sex offenses, robbery, murder and kidnappings. High crime rate makes people feel insecure, thus become one of the major problems in society. In the recent era of increasing volume crimes, crime prevention is now one of the most important global issues, along with the great concern of strengthening public security. The crime prediction method is to predict the crime rates that help police officer to prevent the crime rates in effective way. From that, we propose a new prediction method based on hybrid classifier. The proposed methodology consist of four stages namely, data collection, pre-processing, feature selection and prediction. Initially, the data are collected from internet. Then, in pre-processing stage, redundant and missing values are removed. After that, in feature selection process, important features are selected. Finally, the selected features are given to the hybrid classifier. Hybrid classifier is a combination of support vector machine (SVM) and artificial neural network (ANN). Finally, based on the score value, the crimes are predicted. The performance of proposed methodology is analysed in terms of different metrics. The proposed method implemented in JAVA. From the result, this method efficiently retrieves the features of the crime with a high accuracy of 94.028% compared to other existing methods. |
