Limits of Predictability in Human Mobility

Abstract

Predictable Travel Routines While people rarely perceive their actions to be random, current models of human activity are fundamentally stochastic. Processes that rely on human mobility patterns, like the prediction of new epidemics, traffic engineering, or city planning, could benefit from highly accurate predictive models. To investigate the predictability of human dynamics, Song et al. (p. 1018 ) used the recorded trajectories of millions of mobile phone users, collected by mobile phone companies and anonymized for research purposes. They hypothesized that given the wide range of travel patterns that different users follow, there would be significant differences between their predictability as well: Users who travel less should be easier to predict than those who are constantly on the road. Surprisingly, there was 93% predictability across the whole user base, and individuals' predictability did not in general fall significantly below 80%.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 3,230 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.