Statistical models of neural networks for medical insurance cost forecasting

Abstract

Neural networks have become a popular method of non–linear statistical forecasting therefore it seems important to apply this method to forecasting the cost of health insurance. In this paper the network based on a stochastic model that has direct connection tiered architecture with casual connections between modules and frequency characteristics with obstacles. The resulting Bayesian method, derived by logical solution for this model is based on the Kalman filter. Thus resulting learning algorithm generalizes the so–called one–dimensional Newton method that implements the algorithm is now popular in the literature on neural networks. The article presents numerical prediction method for studying relative cost of health insurance in the form of chaotic time series of errors and shows higher forecasting accuracy of the new algorithm compared to the existing ones.

Authors and Affiliations

A. SIGAYOV

Keywords

Related Articles

Evaluation of resource efficiency of production potential for the criteria "marginal product resource - resource price"

In the article the methodical approaches to determining the structure and presentation of resources, production capacity, efficiency assessment carried out using the criterion of "marginal product of the resource - the p...

Problems and main directions of improvement of balance Ukraine's labor market

The article analyzes problems and directions of improvement of balance the labor market in Ukraine, including the problem of employment and unemployment by age groups, causes unemployment unemployed, demand, supply and e...

Maximization of the banking business value as a key criteria for the assessment of the efficiency of banking management

The article proposes practical mechanisms for improving approaches to managing and assessing the value of banking business from the point of view of a key criterion for assessing the effectiveness of bank management. Adv...

International experience of formation and development of export control systems

The results of generalization of international experience and mechanisms of formation of export controls. Highlight key features of formation and development of export control systems in the US and EU. Proposals to imple...

Problems of tax regulation of the subjects of software product industry in Ukraine

he article reveals the shortcomings of the tax regulation of subjects of software industry in Ukraine. Monitored its effectiveness. Proposed solutions to the problems of tax regulation of subjects of software industry, a...

Download PDF file
  • EP ID EP277415
  • DOI -
  • Views 133
  • Downloads 0

How To Cite

A. SIGAYOV (2017). Statistical models of neural networks for medical insurance cost forecasting. Формування ринкових відносин в Україні. Збірник наукових праць, 1(1), 61-67. https://www.europub.co.uk/articles/-A-277415