Artificial neural networks method and a research on forecasting economics indicators

Today's technology has improved significantly in the storage andprocessing of data to obtain new information. Accordingly, the process ofgenerating information about the new period has gained more importance forpeople based on the knowledge of the previous period. Over time, this case hasbecome an important issue for individuals as well as for countries.In this thesis,the definition of artificial neural networks is explained, its history and generalstructure is explained and an application is made on economicindicators. Theaim of the thesis is to compare the results obtained with the artificial neuralnetwork method with the classical method. The least squares method was used toestimate the economicindicatorsformed from time series."nftool"structure wasused in artificial neural network method.When the results of two differentmethods were examined in the application section, it was found thatthe results ofthe model made with artificial neural networks gave a smaller error.The firstmodel is composed of oneindependent variable and one dependent variable, andthe second model consists of one dependent variable and two independentvariables.It was observed that there was no gap between the results due to thelow number of data and the failure to include other economic indicators in themodels. In today's technology, when all the situations are evaluated, it isconcluded that artificial neuralnetworkmethod ismore useful.