Smart Growth Predictions: Deep Learning Applications in Economic Forecasting

Abstract

Forecasting economic growth of countries is a complex and challenging task, but it has the potential to provide valuable insights for policymakers, investors, and businesses. Several AI based techniques have been used in various time series forecasting applications, including economic growth prediction. This paper explores the potential of Deep Learning (DL) techniques to enhance the accuracy of forecasting by extensively examining data from the CEIC Data Global Database. It focuses on key economic indicators such as Gross Domestic Product (GDP) growth, inflation, unemployment etc. across 50 countries, grouped into five regions, over the past 20 years. The proposed DNN model is developed in python using Keras /TensorFlow, trained for 200 epochs via Adam optimization and MAE loss to optimize model parameters by making 2000 full passes over the training data. The DL model forecasts Asia achieving the highest GDP growth rate amongst the continents through the period, followed by North America, Europe, and others.