人类个体出行行为的统计实证

Empirical Statistics on Individual Human Travel Behavior

统计了欧洲某城市230位居民在6周内的出行活动中的停留时间分布和出行距离分布。首先假设人类出行的停留时间和距离在群体和个体水平上都符合幂律分布,然后用最大似然估计法对这些分布的幂指数进行估计,最后用Kolmogorov-Smirnov检验法对假设的真伪性进行检验。结果发现:在停留时间分布方面,无论在群体水平还是个体水平上都不具有幂律分布特征,而是以相对幂律分布更高的概率进行长时间停留;在出行距离分布方面,群体的出行距离服从带有指数截断的幂律分布,但有198个个体的出行距离分布并不具有幂律特征,而是在某个特征性距离上出现峰值。对这些统计特征的形成机制进行了解释。

Empirical statistics of the waiting time distributions and the travel distance distributions of 230 residents in an European city for a six-week period is presented. On the hypothesis that the waiting time distributions and the travel distance distributions are power-law at both population and individual levels, the maximum likelihood method is used to estimate the parameters of the distributions. And then the Kolmogorov-Smirnov method is used to test the hypothesises. The results show that the waiting time distributions, neither at population level nor at individual level, are power-law. The probability of that an individual staying at a location for a long time is larger than the prediction of the power-law distribution. The results also show that the travel distance distribution of human can be fitted with a truncated power-law at population level. Yet, at individual level, 198 users have peaked travel distance distributions, peaking at the distance between his or her two most visited locations. The mechanisms of these statistical characteristics are exDlained.