Modeling Rideshare Demand: A Bi-Modal Conditional Approach

Ridesharing platforms offer significant potential to address transportation inequities, yet prior research has predominantly focused on urban centers with robust public transit systems, often overlooking suburban regions lacking such infrastructure. This study investigates the ridesharing demand in Arlington, Texas, a suburban area without fixed-route public transportation, over a two-year period. A novel hybrid modeling framework was employed that integrated a primary classification model with a secondary regression model to forecast ride volumes between specific origin-destination census tracts, and a comparative analysis demonstrated that this approach outperformed conventional modeling techniques in predictive accuracy. A feature importance analysis indicated that the time of day and origin and destination tracts were the most influential factors in predicting ride counts, whereas variables such as the month or day of the week were of negligible impact. These findings provide valuable insights for those involved in transportation planning, urban policy development, and future research directions.