Fig. 1: Self-organization of routes in informal public transport networks.

Illustration of the data for the urban region of Cochabamba, Bolivia, South America. a A bus route is determined from data as the trajectory driven by a bus, spatially sampled as a sequence of GPS positions (see “Methods”). b Different routes belonging to the same ID vary from each other only locally. c.i–c.iii, d Several variations of the route self-organize into consistent lines with high overlap. The three routes only differ at the beginning and the end as well as at a small localized part highlighted by pink rectangles in panels (c) and (d) shown in more detail in panels (a) and (b). e The collection of all 431 tracked routes with 130 different IDs covers the whole city of Cochabamba. f, g Counting the number of routes crossing 400m-sized hexagonal areas of the city estimates the service density that resembles the population density, a rough proxy for the local demand (see Supplementary Note 3 for details).