Bootstrap aggregation accuracy gains in nearest‐neighbor nowcasting of Swedish gross domestic product. Issue 2 (16th June 2021)

• Record Type: Journal Article
• Authors: Jönsson, Kristian
• Is Part Of: Applied AI Letters. Volume 2:Issue 2(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-06-16
• Subjects: bootstrap aggregation -- Economic Tendency Survey -- forecasting -- GDP -- machine learning -- nearest neighbor -- nowcasting; 006.3
• DOI: 10.1002/ail2.32
• Languages: English
• ISSNs: 2689-5595

Abstract

Abstract: This article investigates the accuracy gains that can be made by applying bootstrap aggregation to K nearest‐neighbor nowcasting of Swedish gross domestic product. Using both a simulation‐based approach and a theoretical approach, the results indicate that substantial nowcasting accuracy gains can be made using bootstrap aggregation when considering one neighbor in the nowcasting algorithm, that is, when K = 1. Furthermore, a comparison of the simulation‐based and theoretical approaches to bootstrap aggregation indicates that using as few as 25 bootstrap replications can carry, in expectation, the simulation‐based results close to the theoretical results.