Predictive Economic Data Analysis by Statistics and Artificial Intelligence Techniques

Abstract

The paper presents a conceptual approach for synthesis of models for predictive analysis of economic, financial and marketing indicators by statistic and artificial intelligence tools. The target of predictive study are qualitative and quantitative financial indicators for American capital markets. An assessment has been made of the character and degree of the interrelation between the above indicators by correlation analysis. Polynomial mathematical regression models with established high levels of coefficient of certainty at a specific level of significance have been derived. Synthesis of predictive models on the basis of artificial intelligence during Levenberg-Marquardt training, based on the analysis of Mean-Squared Error criterion.