Prominent examples of collective systems are often encountered when analysing smart cities and smart transportation systems. We propose a novel modelling and analysis approach combining statistical model checking, spatio-temporal logics, and simulation. The proposed methodology is applied to modelling and statistical analysis of user behaviour in bike sharing systems. We present a tool-chain that integrates the statistical analysis toolkit MultiVeStA, the spatio-temporal model checker topochecker, and a bike sharing systems simulator based on Markov renewal processes. The obtained tool allows one to estimate, up to a user-specified precision, the likelihood of specific spatio-temporal formulas, such as the formation of clusters of full stations and their temporal evolution.