Forecasting the Gross Domestic Product using LSTM and ARIMA

Abstract

This article aims to predict the Gross Domestic Product (GDP) time series using two different methods: a connectionist approach based on self-learning by a Long Short-Term Memory (LSTM) network, and a statistical approach based on the Box and Jenkins method, represented by an Autoregressive Integrated Moving Average (ARIMA) process. The main objective is to compare the performances of the two approaches in terms of mean squared error (MSE) and determine which method provides more accurate and reliable predictions. Based on the experimental results, the LSTM network outperforms the ARIMA model in predicting the GDP time series. The LSTM network achieved a lower MSE of 0.010 compared to the ARIMA model’s MSE of 0.095. This suggests that the LSTM network achieves better prediction accurcy than the ARIMA model in predicting GDP.