Forecasting demand fluctuations of public bus transit during special events and adverse weather conditions through smart card data analysis

Abstract

The demand for public transportation is influenced by various factors daily, creating significant challenges for managing the fleet. This study aims to examine the demand for bus fleets in Mashhad, Iran, under different weather conditions and special events. Big data from 13 municipal districts collected via smart cards during the one-year period spanning from November 1, 2021, to December 1, 2022, was analyzed. The demand volume was modeled using the seasonal autoregressive integrated moving average (SARIMA) model in conjunction with the Ljung-Box approach. Initially, the data was visualized, and then differential approaches and logarithmic transformations were employed for modeling after identifying the seasonal effect and achieving stationary mean and variance. Extensive smoothing was also utilized to compare the performance of the estimated models. Dynamic regression analysis of time series with SARIMA errors was employed to investigate the impact of temperature, rainfall, snowfall, and special days on passenger demand. The study indicated that demand fluctuations are associated with the district’s outlined land use and cultural and demographic factors under varying weather and special day conditions. Moreover, the findings affirmed that passenger behavior is intricate and localized. After analyzing the factors, it was noted that rainfall impacts the demand for the public transportation system, leading to an 8% decrease. Moreover, this reduction escalates to 37% during snowfall. However, temperature changes have minimal influence and may not merit attention. There is a projected 46% decline in bus service demand among passengers on special occasions.