Understanding Ridesplitting Behavior with Interpretable Machine Learning Models Using Chicago Transportation Network Company Data

Abstract

As congestion levels increase in cities, it is important to analyze people’s choices of different services provided by transportation network companies (TNCs). Using machine learning techniques in conjunction with large TNC data, this paper focuses on uncovering complex relationships underlying ridesplitting market share. A real-world dataset provided by TNCs in Chicago is used in analyzing ridesourcing trips from November 2018 to December 2019 to understand trends in the city. Aggregated origin–destination trip-level characteristics, such as mean cost, mean time, and travel time reliability, are extracted and combined with origin–destination community-level characteristics. Three tree-based algorithms are then utilized to model the market share of ridesplitting trips. The most significant factors are extracted as well as their marginal effect on ridesplitting behavior, using partial dependency plots for interpretation of the machine learning model results. The results suggest that, overall, community-level factors are as or more important than trip-level characteristics. Additionally, the percentage of White people highly affects ridesplitting market share as well as the percentage of bachelor’s degree holders and households with two people residing in them. Travel time reliability and cost variability are also deemed more important than travel time and cost savings. Finally, the potential impact of taxes, crimes, cultural differences, and comfort is discussed in driving the market share, and suggestions are presented for future research and data collection attempts.

Strangers On This Road We Are On: A Literature Review of Pooling in On-Demand Mobility Services

“…Khaloei et al ( 25 ) developed a mixed logit model using data from a stated- and revealed-preference survey conducted in the U.S.A. and found that commuter trips in an autonomous ridehailing environment would likely result in mostly solo ridehailing trips. Abkarian et al ( 26 ) studied Chicago’s TNC trip data from November 2018 to December 2019 and found that shared rides amounted to 20%–30% of all ridesourcing trips throughout the day (implying that ridesplitting choice was not dependent on trip purpose). A study of Didi Chuxing data from November 2016 found that ridesplitting trips were more frequent on weekdays during the middle of the day rather than during the morning or afternoon peak commuting travel period; meanwhile, ridesplitting on weekends was more prevalent for evening trips ( 27 ).…”

“…Most research looking at data on race and ethnicity in ridesourcing tends to show that non-white persons are more likely to choose a ridepool option than non-Hispanic/Latinx white populations. Studies of shared TNC trip data in Chicago from 2019 using advanced statistical models and machine learning techniques to estimate sharing decisions in TNC trips found that census tracts with higher percentages of non-white persons were positively associated with having shared rides occur in those neighborhoods and that the proportion of the white population at the drop-off area was negatively correlated with ridesplitting ( 26 , 29 ). Meanwhile, the 2019 Austin survey analysis using a joint RP-SP model ( 35 ) found that non-Hispanic/Latinx whites had a lower propensity to choose pooled rides.…”

A Ridepooling Bidding Model Considering Service Quality from the Perspective of Cournot Game Theory

“…This model integrates passenger-vehicle matching and vehicle routing optimization. It not only optimizes order allocation, but also takes into account the impact of cost growth and profit reduction caused by passengers' route differences, so it is more practical and significant (23)(24)(25). However, this kind of model usually assumes that all vehicles have a common starting and ending station, which is different from the random spatial position of vehicles in practice.…”

Spatiotemporal Analysis of the Relationships Between Pricing Strategies and Willingness to Share for Ridesourcing Modes

“…Surge pricing is discussed within the classic economic theory on supply and demand dynamics, as well as other factors such as traffic conditions and trip length. For WTS behavior, recent behavior studies have yielded inconsistent conclusions on the factors governing the behavior (9)(10)(11), for example, travel impedance versus sociodemographic features, although they used the same data from the same geography and for the same period to analyze and model the behavior. One can attribute these inconsistent results to the difference in the granularity level of the analysis, that is, community level as opposed to census tract level.…”