Developing a pedestrian route choice model for university students in Kampala : case study : Makerere University

Abstract

Universities in African cities are intensely walkable, yet evidence on how route features shape pedestrian choices is scarce. This study models student route choice at Makerere University in Kampala using both stated preferences from 106 respondents and revealed preferences for 309 trips from 105 participants captured with a mobile GPS application. Choice sets were generated with Yen’s k-shortest paths on the OSM campus network, and the route attributes such as length, greenery, sidewalk availability, slope, and amenity presence were computed in ArcGIS. Multinomial and path-size logit models were estimated using Apollo package in R software. Whereas the SP model indicates disutility for time and slope and positive utilities for greenery and amenities, the RP model revealed that, students frequently choose longer, greener, off-street routes, avoiding busy routes with high amenity counts and sidewalks. Joint SP and RP estimation tightened model inference with five of six coefficients significant at 5% level of confidence and recovered a large SP scaling parameter, consistent with lower SP error variance and efficient experiment design. Out-of-sample validation with repeated 70/30 holdouts showed small estimation–validation gaps with the joint model exhibiting the best generalisation. Policy-wise, greening internal connectors and reducing pedestrian exposure to traffic appear more effective than adding sidewalks along busy corridors. The study’s convenience sample and the controlled campus environment limit external validity, but results provide actionable evidence for institutional hubs and a foundation for larger, city-scale studies enriched with situational variables such as time of day, weather, lighting, safety and volume of traffic.