Intelligent geographical information system for criminology

Many researchers have worked on understanding crime occurrences as crime prevention studies. The best practice for crime prevention is to investigate the cause of crime and to prevent it before it occurs again. Correspondingly, it can be said that crime occurrences are not distributed randomly in space and also they can be affected by various factors. Thus, if factors affecting the crime occurrences are identified, a spatial decision support system could developed for decision makers in Police Departments to make high accuracy crime prevention studies. This research aims predicting the potential risk areas of singular criminal activities before they occur according to examine the factors affecting Theft occurrences. The research was performed on two different scale, regional and local scales, to determine which scale is suitable for predicting potential locations of future Theft incidents with various data from a wide range of data sources. Geographically weighted regression method was used to determine factors affecting Theft incidents. Despite crime can be described better at regional scale, security units which take measures of prevention activities has responsibility areas limited by neighborhood borders. Therefore, spatial prediction was performed at local scale. Prediction was carried out in two different output set by using artificial neural networks and geographical information systems together to make spatial decision support system. One of them was estimating the actual number of Theft incidents. The other one was estimating whether a Theft committed in that location or not. Input variables which obtained from geographically weighted regression results were used in the structure of artificial neural networks. At the end of the process, predicted Theft incidents were explained target values with the accuracy of 32.6 percent better than regression methods. As a result, prediction algorithm was performed 91 percent successfully while predicting whether a Theft committed in that location or not.