국제전자상거래 영향요인 분석: 중력 모형과 머신러닝의 결합

Rapidly growing international e-commerce exhibits significant differences in participants and transaction methods compared to traditional international trade, which has been primarily enterprise-centered. Understanding these differences is essential for effectively utilizing international e-commerce. This study analyzes the factors influencing international e-commerce through panel data analysis utilizing machine learning techniques. The panel data uses Korea's country-specific e-commerce export volume as the dependent variable and includes variables such as GDP, distance, the impact of the Korean Wave (Hallyu), e-commerce infrastructure, and logistics infrastructure in e-commerce exports. Initially, the importance of multiple variables is determined using machine learning, and then a panel data analysis is conducted using the highly significant variables to verify the results. The study found that the influence of the Korean Wave, e-commerce infrastructure, and logistics infrastructure significantly affects e-commerce exports. Furthermore, it was confirmed that the variables identified as important through machine learning also have a statistically significant impact, indicating that machine learning can effectively determine the factors influencing export volume. However, since it is impossible to interpret the direction of the impact, it is suggested to use statistical methods in conjunction to supplement this limitation.