search for


Comparative study of prediction models forpublic bicycle demand in Seoul
Journal of the Korean Data & Information Science Society 2021;32:585-92
Published online May 31, 2021;
© 2021 Korean Data and Information Science Society.

Soah Min1 · Yoonsuh Jung2

2Department of Statistics, Korea University
Correspondence to: Jung’s work has been partially supported by National Research Foundation of Korea (NRF) grants funded by the Korea government(MIST) 2019R1F1A1040515 and 2019R1A4A1028134. This paper is based on Soah Min’s Master thesis.
1 BE Professional, DTI Finance Innovation Unit, LG CNS, Seoul, 07795, Korea.
2 Corresponding author: Associate professor, Department of Statistics, Korea University, Seoul 02841, Korea. E-mail:
Received April 30, 2021; Revised May 12, 2021; Accepted May 17, 2021.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Recently, as an alternative to environmental and transportation issues, the policy of activating the use of public bicycles has been mass-produced and the usage has increased. This paper compares and studies the model for predicting the demand of public bicycles based on the daily rental history of public bicycles in Seoul. VAR model which is extended from univariate autoregressive model to multivariate autoregressive model, the SVR model, which does not require independent assumptions, and the LSTM model, which is a deep learning technique specialized in dependent data are compared. In addition, a time series clustering analysis technique that clusters data according to a dissimilarity measure is used in SVR and VAR models. The performance is compared using RMSE and MAE. The predictive power of the LSTM model is the best, and the next is SVR. VAR model shows lowest predictive power compared to the other models.
Keywords : LSTM, public bicycle policy, support vector regression, time series clustering, vector autoregression.