Classificação de indivíduos segundo comportamento individual relativo a viagens a partir de dados em painel obtidos por smartphones

The characterization of travel behavior is a major issue in the activity-based travel analysis and, generally, is the response variable on travel demand modeling. The individual travel behavior classification can be realized using sectional data such as trip distances, travel mode choice or performed activities. Also, can be done using panel data, such as average values for multiple days or recurrent activities. Panel data are an important tool in behavioral analysis related to urban trips, providing extra dimension of analysis related to the individual temporal heterogeneity. However, obtaining these data is not trivial, requiring monetary and time resources. Thus, the main goal of this study is to classify individuals according to travel behavior from panel data. The secondary goal is related with panel data collection through smartphones. The potential of the study is validated by a case study with undergraduate and PhD students from Sao Carlos - SP, Brazil. With data voluntarily provided by the students, a k-means algorithm was employed considering, as input, four variables associated with trips carried out in three consecutive working days. Three different behavioral groups were obtained with differences concerning degree of motorization, recurrence of localities, number of trips performed, and average distances traveled.