engleski

Review on doctoral dissertation

The main issue of the dissertation is of a methodological character, referring to different approaches and econometric models in the analysis of time series that have so far been most frequently used in the measurement and forecasting of the expected inflation. In this dissertation, by systematizing the advantages and disadvantages of traditional econometric models, including different parametric and nonparametric models, and linear and nonlinear models, the author comes to the conclusion of the necessity of application of a semi-parametric approach for measurement and forecasting of the expected inflation, i.e. neural networks.

Tea Poklepović, Neural Network Structure Identification in Measurement of Expected Inflation

nije evidentirano