Pattern recognition based on statistics and structural equation models in multi-dimensional data warehouses of social behavioral data

Abstract

The approach of using statistical analysis and structural equation modeling methods in support of multiple analyses and semi-automatic pattern recognition processes is proposed. The multidimensional decision support system is integrated with the data mining technologies of social data warehouses. Ensembles of diverse and accurate classifiers are constructed on the bases of multidimensional classification methods and allow more sophisticated relations between variables of data analysis and factor analysis to be revealed. The paper describes an approach in which linear equation models are integrated with multiple statistical analysis and knowledge representation to recognize information patterns in data warehouses. The introduced methods of analysis allow us to practice control and to forecast the main social tendencies, as well as supporting decisions in different prevention tasks. The results of the application of the methods of statistical analysis are demonstrated by the social behavioral analysis data (for example, criminal analysis) of the some EU Countries.