DOI QR코드

DOI QR Code

KTX passenger demand forecast with multiple intervention seasonal ARIMA models

다중개입 계절형 ARIMA 모형을 이용한 KTX 수송수요 예측

  • Cha, Hyoyoung (Department of Statistics, Hankuk University of Foreign Studies) ;
  • Oh, Yoonsik (Research Center, Korea Railroad) ;
  • Song, Jiwoo (Department of Statistics, Hankuk University of Foreign Studies) ;
  • Lee, Taewook (Department of Statistics, Hankuk University of Foreign Studies)
  • 차효영 (한국외국어대학교 통계학과) ;
  • 오윤식 (한국철도공사연구원) ;
  • 송지우 (한국외국어대학교 통계학과) ;
  • 이태욱 (한국외국어대학교 통계학과)
  • Received : 2018.11.15
  • Accepted : 2018.12.08
  • Published : 2019.02.28

Abstract

This study proposed a multiple intervention time series model to predict KTX passenger demand. In order to revise the research of Kim and Kim (Korean Society for Railway, 14, 470-476, 2011) considering only the intervention of the second phase of Gyeong-bu before November of 2011, we adopted multiple intervention seasonal ARIMA models to model the time series data with additional interventions which occurred after November of 2011. Through the data analysis, it was confirmed that the effects of various interventions such as Gyeong-bu and Ho-nam 2 phase, outbreak of MERS and national holidays, which affected the KTX transportation demand, are successfully explained and the prediction accuracy could be quite improved significantly.

본 연구는 KTX 수송수요를 예측하기 위한 방법으로 다중개입 시계열 모형을 제안하였다. 구체적으로 2011년 이전의 자료로서 경부 2단계 개통 개입만 고려한 Kim과 Kim (Korean Society for Railway, 14, 470-476, 2011)의 연구를 수정 보완하기 위해 다양한 개입이 추가적으로 발생하고 있는 2011년 이후의 시계열 자료를 효과적으로 모델링하는 한편 KTX 수송수요를 정확히 예측하기 위한 방법으로 다중개입 계절형 ARIMA 모형을 도입하였다. 자료 분석을 통해 KTX 수송수요에 영향을 주었던 경부 및 호남 2단계 개통, 메르스 발병과 설추석 명절 등 다양한 개입의 효과를 효과적으로 해석하는 한편, 이를 통해 예측의 정확성을 높일 수 있음을 확인하였다.

GCGHDE_2019_v32n1_139_f0001.png 이미지

Figure 2.1. Gyungbu line (a) and Honam line (b) monthly average daily trips.

GCGHDE_2019_v32n1_139_f0002.png 이미지

Figure 3.1. Plot of residual ACF.

Table 3.1. Fitted models for Gyeong-bu and Ho-nam KTX Lines based on AIC

GCGHDE_2019_v32n1_139_t0001.png 이미지

Table 3.2. Residual analysis of multiple intervention seasonal ARIMA models for Gyeong-bu and Ho-nam line

GCGHDE_2019_v32n1_139_t0002.png 이미지

Table 3.3. Performances of multiple intervention seasonal ARIMA models for Gyoung-bu and Ho-nam Line

GCGHDE_2019_v32n1_139_t0003.png 이미지

Acknowledgement

Supported by : 한국외국어대학교

References

  1. Box, G. E. P. and Tiao, G. C. (1975). Intervention analysis with application to economic and environmental problems, Journal of the American Statistical Association, 70, 70-79. https://doi.org/10.1080/01621459.1975.10480264
  2. Cho, S., Sohn, Y. and Seong, B (2016). Time Series Analysis using SAS/ETS, Yulgok Books, Seoul.
  3. Choi, T. and Kim, S. (2004). An empirical comparison among initialization methods of Holt-Winters model for railway passenger demand forecast, Journal of the Korean Society for Railway, 7, 9-13.
  4. Cryer, J. D. and Chan, K. S. (2008). Time Series Analysis: With Applications in R, Springer-Verlag, New York.
  5. Hong, C. and Lee, G. (2016). Study of factors affecting the size of government employment in Korea, Korean Public Personnel Administration Review, 15, 29-60.
  6. Huh, H. J. and Kim, H. C. (2001). Forecasting demand for Jeju-bound tourist: an application of intervention method, Journal of Tourism Sciences, 25, 27-42.
  7. Kim, K. and Kim, H. (2011). KTX passenger demand forecast with intervention ARIMA model, Korean Society for Railway, 14, 470-476. https://doi.org/10.7782/JKSR.2011.14.5.470
  8. Kim, S. and Seong, B. (2011). Intervention analysis of Korea tourism data, Korean Journal of Applied Statistics, 24, 735-743. https://doi.org/10.5351/KJAS.2011.24.5.735
  9. Kim, Y. and Jo, J. (2014). A study on demand forecasting for KTX passengers by using time series models, Korean Journal of Applied Statistics, 27, 1257-1268. https://doi.org/10.5351/KJAS.2014.27.7.1257
  10. Song, J. (2016). A study on demand forecasting for Jeju-bound tourists by travel purpose using seasonal ARIMA-intervention, Journal of the Korean Data & Information Science Society, 27, 725-732. https://doi.org/10.7465/jkdi.2016.27.3.725