Vehicle travel path recognition in urban dense road network environments by using mobile phone data

Activity‐based model based on multi‐day cellular data: Considering the lack of personal attributes and activity type

“…where P denotes the set of density‐connected clusters, i.e., alternative activity locations; $${P}_l$$ denotes the l th alternative location; $${p}_o$$, $${p}_u$$, and $${p}_j$$ denote the o th, u th, and j th POI data in $${P}_l$$, respectively; k denotes the number of POI data in $${P}_l$$; $$d( {{p}_o,{p}_u} )$$ denotes the distance between$${p}_o$$ and $${p}_u$$; $$\Delta d$$ denotes the spatial radius constraint; and $$\Delta l$$ denotes the number constraint. The core POIs and cluster extension conditions were determined jointly using the two types of constraint parameters [23].…”

“…Mobile phone data, as a passive data acquisition technique, records individual travel information for a long period and in great detail through GPS or cellular communication traces. Such data has been widely used for travel features recognition, such as travel OD, travel mode, travel path, and transferring [7,[21][22][23]. Therefore, it can provide good basis for ABM construction [24,25].…”