A Generalized Additive Logit Model of Brand Choice

The paper presents an econometric application of generalized additive Logit regression models (GALRMs) for brand choice. Our semi-parametric models are flexible and robust extensions of the Logit model. The GALRMs are fit to binary response data by maximizing a penalized log likelihood or a penalized log partial-likelihood. The GAMs allow us to build a regression surface as a sum of lower-dimensional nonparametric terms circumventing the curse of dimensionality: the slow convergence of an estimator to the true value in high dimensions. Four GALRMs are compared with a Logit model for brand choice and the best model is selected using various model selection criteria.