Hybrid Models Application for Short Time Series Prediction Based on Social and Economic Indicators of Russian Federation

Abstract

The key to the successful operation of companies in today's economy is the rapid decision-making process based on precise forecasts. However, key indicators do not usually have sufficient quality and quantity of source data, which leads to failure of standard methods and the prediction models in some areas of the economy. This main objective of this paper is to analyze the models for short series prediction that are based on multiple linear regression and neural networks (with two different learning algorithms: Back propagation and BFGS) on the example of socio-economic indicators in the Russian Federation.