Mutualistic networks arise out of several ecological processes including pollination, seed dispersal, and host-parasite relationships. Studies on these networks have revealed common structural patterns, such as nestedness, but few have quantifi ed the extent to which functional traits affect the probability of, say, plants and pollinators interacting with each other. We took an econometric approach that treated each pollinator as a consumer faced with several choices (corresponding to the plant species in the network). In particular, we used grouped Dirichlet-multinomial regression to model the interaction probabilities of plant-pollinator networks as a function of plant and pollinator traits. In this talk, we present our regularized grouped Dirichlet-multinomial regression to facilitate identification of plant and pollinator characteristics predictive of interaction when a lot of trait information is available.