In analysis of macroeconomic indicators, the most familiar statistical method is OLS if the underlying model is linear. However, in the age of “big data”, datasets with a large dimension are becoming more common so that the OLS regression method may fail to estimate the parameter coefficient and significant subset selection. Therefore, shrinkage algorithms such as RIDGE regression and LASSO regression are increasingly becoming an important tool in statistics due to its freedom of OLS assumptions, especially in time series analysis. Indeed, LASSO regression provides efficient and fast results based on its shrinkage technique, especially in the case of highly dimensional and correlated explanatory variables. Then, an illustration is shown using real time series data on Vietnam’s export. We practice model estimation on both OLS and LASSO regression and point out that LASSO regression is an effective alternative to OLS regression in time series prediction. Specifically, in this study, in comparison with the results predicted by the OLS regression, the LASSO-errors reduced by about 3%.
CITATION STYLE
Uyen, P. H., Uyen, V. T. L., Hoa, L. T., & Trung, T. Q. (2022). LASSO Regression and Its Application in Forecasting Macro Economic Indicators: A Study on Vietnam’s Exports. In Studies in Computational Intelligence (Vol. 983, pp. 575–585). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-77094-5_44
Mendeley helps you to discover research relevant for your work.