Selecting a model for trend-based criminological forecasting

Introduction: the article is devoted to substantiating the choice of a model for criminological forecasting based on the trend of a time series. It is noted that in the absence of a scientifically based mathematical form for predicting crime in each specific case of criminological forecasting, one has to choose one of the possible analytical dependencies. The article examines the issues of assessing the quality of models and their suitability for forecasting. Materials and methods: the research is based on mathematical methods of time series analysis. Findings and Conclusions: a technology for analyzing mathematical models is proposed, which allows choosing the best model for criminological forecasting based on trend in terms of forecast accuracy. The analysis of the model includes three stages: assessment of the quality of the model, assessment of the suitability of the obtained model for forecasting and assessment of the forecast error. The proposed approach is implemented to solve the problem of short - term forecasting of the number of registered crimes in the Irkutsk region.