Modelling the Gross Domestic Product and the Per Capita Income of Türkiye using Autoregressive Deep Learning Networks

Abstract

The gross domestic product and the per capita income are leading indicators regarding the income and the wealth of nations. The gross domestic product and the per capita income can be modelled dependent on various econometric data such as the export revenue, tourism revenue, trade deficit and the industrial revenue. In this work, an alternative and novel method is presented for the modelling of the gross domestic product and the per capita income. In this study, the gross domestic product and the per capita income are modelled autoregressively employing deep learning networks namely autoregressive deep learning networks. The input data of the developed deep learning networks are taken as the past values of the modelled variable making the deep learning networks effectively autoregressive models. As application examples of the autoregressive deep learning models, the gross domestic product and the per capita income data of Türkiye for the period of 1960-2021 are separately modelled. The autoregressive deep learning networks are developed in Python programming language. The coefficient of determination (R2), mean absolute error (MAE), mean absolute percentage error (MAPE) and the root mean square error (RMSE) of the developed models are also computed. The plots of the results of the developed autoregressive deep learning models and the performance metrics of the models show that the developed autoregressive deep learning models can be utilized to accurately model the gross domestic product and the per capita income.