2025년도 KTX 수요 예측 및 정책적 의사결정: XAI 기반 실증적 예측연구 = Demand Forecast and Policy Decisions for KTX in 2025: An Empirical Forecasting Study based on XAI

다국어 초록 (Multilingual Abstract)

Accurate forecasting of KTX demand is a key factor for improving operational efficiency, establishing transport policies, and building sustainable infrastructure. The main purpose of this study is to accurately forecast KTX demand in 2025 using explainable artificial intelligence and support practical decision-making. Using AI algorithms, we improved accuracy by reducing forecast errors by up to 2.49%. In addition, by visualising the contribution of variables using the SHAP algorithm, we increased the level of confidence in the decision-making process for business policies and resource allocation. The contribution of variables can vary depending on the KTX lines, confirming the need for a business analytics platform that can be utilised in real-time. In 2025, KTX demand may decrease by up to 9.45% compared to last year, but it is expected to increase and stable gradually, rising up to 18.47% compared to pre-COVID-19.