Forecasting impacts of covid-19 on Vietnam industrial sector: Machine learning approach

Abstract

The objective of this study is to propose one or multiple good performance time-series forecasting models to forecast the Industrial Production Index (IPI) value of Vietnam in the short and medium term under the conditions of the COVID-19 pandemic worldwide predictions based on historical data. The more accurate IPI forecasting, the more effective the government's policy can help both the economic and social well-being of Vietnam be less affected by the pandemic. Our finding is that machine learning is better than traditional models and multivariate is better than univariate. The best model performance is Neural Prophet, which is used to forecast the IPI value in the next twelve months. The result is positive when Vietnam's IPI value will experience an upward fluctuation. That is, Vietnam's industrial production is limited by the negative effects of the pandemic and continues to grow. Multi-linear Regression and Sarima are also good alternatives in terms of convenience and saving.