Answers are grounded in this paper and cite their sources.
Randomized trial evaluates mode choice prediction in varied transport options, suggesting enhanced predictability.
The paper introduces Stochastic Encoder Nested Logit (SENL), an ANN-based framework for predicting mode choice using random utility maximization concept. The model preserves useful interpretable insights while enhancing predictability through stochastic randomization of encoder latent space. The model allows for varied representations of apparently alike individuals while also relaxes the IIA property of MNL using nesting structures. This allows us to combine the interpretability of discrete choice models, particularly nested logit model, and the flexibility of deep learning techniques. The SENL model surpasses proposed benchmark models in predictive capability while being tested on a real-world dataset. The paper compares two variants of the SENL model to understand the impact of origin-destination variables in mode choice analysis. The log-sum parameter was found to be 0.652 for the superior variant, thus reinforcing the theoretical plausibility of the model.
No takes yet. Share an insight, caveat, or question.
Panda et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: