Capturing Heterogeneity in the Multinomial Logit Model by Implementing the Bagging Method

摘要: When a multinomial logit model (MNL) is constructed by selecting the best (e.g. highest t-values) subset of independent variables using Maximum Likelihood (ML) approach stability parameters not guaranteed and has been discussed in literature. The definition instability Breiman (1) implies that when unstable small change its train dataset results considerable changes structure model. Thereby, biased prediction error. bagging method, i.e. utilizing an ensemble models instead single introduced literature to be effective reducing for some modelling formulations. Bagging can also increase overall model’s goodness-of-fit. This paper investigates effectiveness implementing method MNL. It discusses required condition where MNLs, called Random MNL (RMNL), improves log-likelihood value compared Furthermore, capability RMNL capturing taste variation explained accuracy against mixed (MMNL) as well-known addressing heterogeneity. A publicly available “college distance” used demonstrate impacts on