In this paper we propose a general framework for pedestrian walking behavior, based on discrete choice modeling. Two main behaviors are identified: emphunconstrained and emphconstrained. The constrained patterns are further classified into emphattractive interactions and emphrepulsive interactions. The formers are captured by a emphleader-follower model while the latters through a emphcollision avoidance model. The spatial correlation between the alternatives is taken into account defining a cross nested logit model. Quantitative analysis is performed by maximum likelihood estimation on a real dataset of pedestrian trajectories, manually tracked from video sequences.