We introduce a semi-supervised discrete choice model with algorithmic approaches to estimate choice models when relatively few requests have actual preferences but the majority only have the choice sets. Two classic semi-supervised learning algorithms, the expectation maximization algorithm and the cluster-and-label algorithm, have been adapted to our choice modeling problem setting. We also develop two new algorithms based on the cluster-and-label algorithm. The new algorithms use the Bayesian Information Criterion to evaluate a clustering setting to automatically generate new clusters out of existing clusters and adjust the number of clusters. Two computational studies focusing on travel demand forecasting (i.e., a hotel booking case and a large-scale airline itinerary shopping case) are presented to evaluate the prediction accuracy and computational effort of the proposed algorithms. Algorithmic recommendations are rendered under various scenarios based on the hotel booking case while economic insights are derived based on the itinerary shopping case.
Semi-Supervised Learning for Discrete Choice Models
IEEE Transactions on Intelligent Transportation Systems ; 20 , 11 ; 4145-4159
2019-11-01
2686792 byte
Article (Journal)
Electronic Resource
English
Enhancing discrete choice models with representation learning
Elsevier | 2020
|Preference endogeneity in discrete choice models
Online Contents | 2014
|Bayesian Reliability of Discrete Choice Models
British Library Conference Proceedings | 1998
|Dynamic Discrete Choice Models for Transportation
Online Contents | 2011
|