Artificial neural network as an alternative to multiple regression analysis for estimating the parameters of econometric models
¹Institute of Economics and Social Sciences, Estonian University of Life Sciences, Kreutzwaldi 64, 51014 Tartu, Estonia; e-mail: Reet.Poldaru@emu.ee, Jyri.Roots@emu.ee
²ARIB, Narva 3, 51009, Tartu, Estonia; e-mail: Ants.Viira@pria.ee
In recent years, neural networks have been used for a wide variety of applications where statistical methods are traditionally employed. Neural nets offer the opportunity to create a model by using technology similar to the learning patterns of the human brain. The structure of artificial neural networks (ANN) is based on the human brain’s biological neural processes. Artificial neural networks provide a new approach to the problem of parameter estimation of nonlinear econometric models. This paper presents a comparison between neural networks and econometric approaches for estimation of parameters of an econometric model of grain yield. The aim of this study is to show that neural nets are a convenient econometric tool. The parameters were estimated on the basis of alternative variants of models. The analysis shows that artificial neural network models may be used for parameter estimation of the econometric models.