It has been recently claimed that human movement is ‘highly predictable’ upper bound of 93% predictability shown. However, knowing an upper bound is only useful if it is relatively tight, i.e. it is close to the true limit of predictability. In this work we reconsider the derivation of the upper bound to movement predictability and by considering real-world topological constraints we are able to achieve a tighter upper bound. Our results show that the upper bound is actually between 11-24% less than previously claimed.
Understanding upper bounds on how predictable human travel is important for both understanding and planning for human movement in both urban and urban environments. Additionally such bounds enable designers of applications and services that rely on accurate prediction of human movement to understand and work around the constraints with which they are faced. Examples of such applications include ubiquitous advertising, service provision and intelligent agents both virtual and physical.
It has been recently claimed that human movement is ‘highly predictable’ with an upper bound of 93% predictability shown. However, knowing an upper bound is only useful if it is relatively tight, i.e. it is close to the true limit of predictability. This project focused on improving the previous bounds providing a tighter, more accurate alternative.
Outcome and Impact
The project reconsidered the mathematical derivation of the upper bound to human movement predictability. By considering real-world topological constraints an approach to computing a tighter upper bound was developed. The results show that the upper bound is actually between 11-24% less than previously claimed..
Particularly at fine-grained spatial and temporal quantization, where a significant number of practical applications lie, these new (lower) upper limits raise serious questions about the use of location information alone for prediction, contributing more evidence that such prediction must integrate external variables in order to satisfy the prediction accuracy requirement of a significant number of pervasive computing (and other) applications.
An evaluation of existing predictions methods compared to the derived upper bound would provide interesting evidence to how close the upper bound is within datasets with generalizable properties to real world use cases (i.e. evaluation across a number of cities). Finally, the mathematical assumption of stationarity (used in this and prior work) is unlikely to hold (to varying degrees) in reality leading to potential improvements to the approach if such an assumption could be removed.
The work was presented at IEEE International Conference on Pervasive Computing and Communications (PERCOM 2014).
A supporting website, including instructions and code to replicate the results is available.
The slides from the presentation at PERCOM 2014: