Treelet Decomposition of Mobile Phone Data for Deriving City Usage and Mobility Pattern in the Milan Urban Region

Statistical analysis of complex and spatially dependent data: A review of Object Oriented Spatial Statistics

“…BV has been originally conceived to cluster functional data observed over a lattice D 0 . Nevertheless, the algorithm can be applied without substantial modifications to general object data, and for a variety of target statistical analyses (e.g., dimensionality reduction or joint clustering and alignment of data Secchi et al, 2015;Manfredini et al, 2015;Abramowicz et al, 2016). For ease of exposition, we here focus on an Euclidean spatial domain D 0 , although more complex situations may be considered as well; the reader is referred to Cressie (1993, Part II) for an introduction to classical spatial statistics for lattice data.…”

Mapping Urban Practices Through Mobile Phone Data

“…With the Treelet decomposition methodology (Manfredini et al 2012a, b;Vantini et al 2012;Manfredini et al 2015) it is possible to obtain: a reference basis reporting the specific effect of some activities on Erlang data; a set of maps showing the contribution of each activity to the local Erlang signal. In doing so, the Treelet decomposition basis contains different temporal patterns of mobile phone activity (i.e.…”

Discovering Regularity Patterns of Mobility Practices through Mobile Phone Data

“…The methodology of Treelet decomposition (Manfredini et al, 2012;Vantini, Vitelli, & Zanini, 2012) allowed us to obtain: a reference basis reporting the specific effect of some activities on Erlang data; a set of maps showing the contribution of each activity to the local Erlang signal. The idea behind our approach is that different basic profiles (each being one element of the treelet decomposition basis) of city usages can concur in the same place and that the overall observed usage of a certain place is the superimposition of layers of these profiles.…”