Unlike particular comparisons of alternative prediction models on different datasets, Song et al.’s approach is much more fundamental: it does not focus on any specific prediction technique but rather on human behavior, as captured by the available data. This maximum is referred to as the predictability of the person’s mobility. Specifically, given a dataset containing a sequence of locations visited by an individual and representing a sample of the individual’s mobility, they proposed a technique that measures the maximum theoretical accuracy that an ideal prediction model could achieve on that dataset. , in which the authors proposed a fundamental technique that explores the concept of entropy to estimate how predictable a person’s mobility is, as expressed in a given dataset. Answering this question was the focus of a seminal paper by Song et al. Given the difficulties involved in predicting mobility-related behavior, one poses the question of to which extent such behavior can be predicted. However, human mobility is hard to predict, as there are many factors, such as the person’s mood, traffic conditions, and current weather, that play a role in mobility-related decisions. Many previous studies proposed mobility prediction strategies that use a myriad of techniques (e.g., Markov chains, logistic regression, neural networks, and so on), and used different types of data sources (call detail records from mobile traffic, GPS traces, and social media data, among others). Human mobility prediction has broad and important applications in areas such as urban planning, traffic engineering, epidemiology, recommender systems, and advertisement, to name a few. Our experiments show that our metrics are able to capture most of the variability in one’s routine (adjusted \(R^\) of up to 84.9% and 96.0% on a GPS and CDR datasets, respectively), and that routine behavior can be largely explained by three types of patterns: (i) stationary patterns, in which a person stays in her current location for a given time period, (ii) regular visits, in which people visit a few preferred locations with occasional visits to other places, and (iii) diversity of trajectories, in which people change the order in which they visit certain locations. Finally, we rely on previously proposed metrics, as well as a newly proposed one, to understand what affects the predictability of a person’s routine. To that end, we propose a technique that allows us to (i) quantify the effect of novelty on predictability, and (ii) gauge how much one’s routine deviates from a reference routine that is completely predictable, therefore estimating the amount of hard-to-predict behavior in one’s routine. Viewing one’s mobility in terms of these two components allows us to identify important patterns about the predictability of one’s mobility.Īdditionally, we argue that mobility behavior in the novelty component is hard to predict if we rely on the history of visited locations (as the predictability technique does), and therefore we here focus on analyzing what affects the predictability of one’s routine. In this paper, we propose to study predictability in terms of two components of human mobility: routine and novelty, where routine is related to preferential returns, and novelty is related to exploration. Although useful in several scenarios, this technique focused on human mobility as a monolithic entity, which poses challenges to understanding different types of behavior that may be hard to predict. In the case of mobility-related behavior, there exists a fundamental technique to estimate the predictability of an individual’s mobility, as expressed in a given dataset. Given the difficulties in predicting human behavior, one may wish to establish bounds on our ability to accurately perform such predictions.
0 Comments
Leave a Reply. |