Search

An empirical analysis of the login datum of fitness club membership over the past 27 months is implemented, which shows that the interval time distribution of their logins presents obvious power-law properties whether all members, groups or individuals. And the scaling indexes of groups are different and linked with their level of activity. Studies on a certain time further reveal that the statistical characteristics of the same community can not be affected by the length of time. Moreover, the power exponents of individuals are also diverse. A conclusion is given to show that the scaling indexes of interval time distribution is dynamically influenced by the level of activity. Finally, by introducing the interest factor and time threshold parameters into the model, human dynamics model based on habit are obtained in order to explore the production mechanism of exercise behaviour of individuals.

Current empirical results of human dynamics based on temporal characteristics are discussed. By collecting communications data from Sui Gu and Van Gogh, the inter-event time distribution is found to fit a power law with exponent 1.5, which is in accordance with Barabasi’s conclusion. Moreover, it shows that the inter-event time distribution about two men’s letters for five consecutive years also follows power law distribution with the same exponent. Finally, some simulations are given to show the effectiveness.