Classification and regression tree, principal components analysis and multiple linear regression to summarize data and understand travel behavior

Through the combined application of Classification and Regression Tree (CART), Principal Components Analysis (PCA) and Multiple Linear Regression (MLR) it is expected to obtain numeric vari-ables for the application of the PCA, synthesize the original database and a set of variables, and find relations between the new variables (components) and travel behavior. The analysis was based on the origin-destination home-interview survey carried out by METRO-SP in the Sao Paulo Metropolitan Area, in 1997. After observing the results of CART and PCA applications, the authors suggested a taxonomy in ten components that are: (1) Family head characteristics; (2) High income individuals group; (3) Land use; (4) College students; (5) Spouse - woman; (6) Autonomous workers; (7) Level of education; (8) Services sector workers; (9) Home maintenance workers; (10) Familiar characteris-tics. Through the linear models travel behavior is strongly correlated, as expected, with these components. Family heads tend to make home-work based tours as well as use cars more often for worktrips, high income people are more likely to make motorized trips, land use characteristics mainly influence the travel distance ;sequence, women - spouses are more inclined to do non-work trips and autonomous workers usually make shorter worktrips.