Methods in committing serial crimes differ but not entirely random. By generating “Geographical profile”, we tend to assist in crime investigation. By adopting model based on distance-decay principle using CGT and modified Centro graph principle, the article adopts analytical hierarchy process (AHP) to determine the living area of criminals and certifies the accuracy of the model through Peter Sutcliffe case. Eventually, it indicates the prediction coincides with the actual situation well. Based on the criminal mentality, the article predicts the possibility of next crime scene from space tendency (both the distance between the living area of criminal and the crime scene and the adjacent crime scene) .As the prediction is far away from being admitted, we choose another perspective to generate a transition density model to solve the same problem in a more effective way. The model indicates the Bayesian process. With the estimate of the parameters, we gain some useful information to forecast the future crime based on the past criminal locations and the time. In the end, the article verifies the efficiency of the model through a specific situation.