Analysis Of Crime Patterns Based On GIS

With the rapid expansion and development of cities in China, urban population increases rapidly, and criminal behavior has become an important factor affecting the safety of urban residents. Therefore, it has important practical value to further study the characteristics of criminal cases. On the basis of summarizing current crime research at home and abroad, take Changning District’s real crime data as an example, do space-time distribution patterns and prediction research of crime based on GIS technology, to reveal the space-time distribution regularity of the two crime types, and predict the developing trend of crime at a certain space-time scale. The main research results are as follows:(1) Through statistical analysis of the crime data at various time scales, it could be respectively concluded the two kinds of cases’ distribution regularity in different time scales, such as every month, every day in a week, a day of each period and moment. As well as the weather factors and holidays’ overall impact on the number of crime.(2) Using the Adjacent Index and Ripley’K statistics analysis methods to identify the spatial distribution patterns, the results show that these two types of cases are spatially clustered, and crime hot spots exist. On the other hand, Kernel Density Estimation is a non-parametric estimation method based on the characteristics of data itself to research the specific distribution. In this paper, using Kernel Density Estimation method to visualize the spatial distribution of crime spots. A year is divided into four parts according to the seasons. Combined with the characteristics of each season, the spatial distribution and change tendency could be concluded through analyzing these two kinds of cases in each season.(3) Association Rules is used to explore the correlation between various items in the same Transaction Database, the paper chooses the classic Apriori algorithm for analyzing space-time correlation of the two types of cases. On the time scale, every 8 hours is divided into a period of time, so a day is divided into three time periods: ’07-14’,’15-22’,’23-06’. On the spatial scale, the whole region is divided into grids of 500m* 500m, covering the entire study area. Using Association Rules to find the relationships between crime types, time and space, then do space-time crime prediction according to the strong rules which meet the minimum support, confidence and lift threshold. Predict which period of time and grid areas are more prone to occur certain types of crime. Finally, its prediction effect is validated with the actual crime data, and compared with statistical predictions based on GIS software. The results show that the effect of prediction using Association Rules is better than using statistical analysis.The research results show that, through analysis and prediction research of crime on certain space-time scales, it could provide some suggestions for the relevant departments to develop police deployment and patrol scheme, and also provide certain reference for domestic criminal research on specific spatial and temporal scales.