코로나19 확산에 따른 대중교통 이용량의 예측 모형 구축

다국어 초록 (Multilingual Abstract)

The outbreak of COVID-19 significantly changed peoples’ lifestyles. also contributing to a change in their travel behavior. Since public transit usage has decreased due to the fear of infection, improving the passenger flow prediction will help public transportation companies better manage this situation. This study analyzes the prediction of public transportation flow made in South Korea during COVID-19 and proposes a suitable model for prediction. Four models were constructed to predict passenger ridership, and the model with the lowest evaluation metrics value for each city was selected. During the model constructing process, it was found appropriate to predict a day with the previous 14 days, and the prediction results showed that the RNN model performed better than the other models in most of the cities. The prediction results of the RNN model were much better in areas other than the metropolitan areas, while the LSTM model performed better in metropolitan areas. The result shows that the RNN model would perform better at predicting public transportation usage in a short-term pandemic situation such as COVID-19. This research is expected to help companies make better decisions pertaining to public transportation operations, such as scheduling and adjustment of dispatch intervals, during pandemic situations.