Abstract
Aggregated mobility indices (AMIs) derived from information and communications technologies have recently emerged as a new data source for transport planners, with particular value during periods of major disturbances or when other sources of mobility data are scarce. Particularly, indices estimated on the aggregate user concentration in public transport (PT) hubs based on GPS of smartphones, or the number of PT navigation queries in smartphone applications have been used as proxies for the temporal changes in PT aggregate demand levels. Despite the popularity of these indices, it remains largely untested whether they can provide a reasonable characterisation of actual PT ridership changes. This study aims to address this research gap by investigating the reliability of using AMIs for inferring PT ridership changes by offering the first rigorous benchmarking between them and ridership data derived from smart card validations and tickets. For the comparison, we use monthly and daily ridership data from 12 cities worldwide and two AMIs shared globally by Google and Apple during periods of major change in 2020-22. We also explore the complementary role of AMIs on traditional ridership data. The comparative analysis revealed that the index based on human mobility (Google) exhibited a notable alignment with the trends reported by ridership data and performed better than the one based on PT queries (Apple). Our results differ from previous studies by showing that AMIs performed considerably better for similar periods. This finding highlights the huge relevance of dealing with methodological differences in datasets before comparing. Moreover, we demonstrated that AMIs can also complement data from smart card records when ticketing is missing or of doubtful quality. The outcomes of this study are particularly relevant for cities of developing countries, which usually have limited data to analyse their PT ridership, and AMIs may offer an attractive alternative.