Application of machine learning methods in forecasting economic growth and inflation of Vietnam

Abstract

Inflation and economic growth are two crucial indicators for any country in the world. In light of the importance of these two economic indicators, the forecast of economic growth and inflation has become a significant topic that national governments have traditionally prioritized. This study aims to apply popular machine learning algorithms such as KNN and MLP to build models for predicting economic growth and inflation. We also provide a comparison of the predictive accuracy between these machine learning algorithms and traditional forecasting models such as VAR and LASSO. Specifically, we employ techniques such as VAR, LASSO, KNN, and multi-layer perceptron (MLP) to construct forecasting models for Vietnam’s economic growth and inflation using data collected from 1996 to 2021. The accuracy of the models is assessed using three indices: RMSE, MAE, and MSE. The empirical results show that according to all three indicators, RMSE, MAE, and MSE, the forecasting models of economic growth and inflation by the MLP model are the most accurate. Based on the results, we have concluded that the MLP model is a valuable tool for future forecasting because it can describe the nonlinear relationships between variables in the model and visually map them.