English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Discrete choice models (DCMs) require a priori knowledge of the utilityfunctions, especially how tastes vary across individuals. Utilitymisspecification may lead to biased estimates, inaccurate interpretations andlimited predictability. In this paper, we utilize a neural network to learntaste representation. Our formulation consists of two modules: a neural network(TasteNet) that learns taste parameters (e.g., time coefficient) as flexiblefunctions of individual characteristics; and a multinomial logit (MNL) modelwith utility functions defined with expert knowledge. Taste parameters learnedby the neural network are fed into the choice model and link the two modules.  Our approach extends the L-MNL model (Sifringer et al., 2020) by allowing theneural network to learn the interactions between individual characteristics andalternative attributes. Moreover, we formalize and strengthen theinterpretability condition - requiring realistic estimates of behaviorindicators (e.g., value-of-time, elasticity) at the disaggregated level, whichis crucial for a model to be suitable for scenario analysis and policydecisions. Through a unique network architecture and parameter transformation,we incorporate prior knowledge and guide the neural network to output realisticbehavior indicators at the disaggregated level. We show that TasteNet-MNLreaches the ground-truth model's predictability and recovers the nonlineartaste functions on synthetic data. Its estimated value-of-time and choiceelasticities at the individual level are close to the ground truth. On apublicly available Swissmetro dataset, TasteNet-MNL outperforms benchmarkingMNLs and Mixed Logit model's predictability. It learns a broader spectrum oftaste variations within the population and suggests a higher averagevalue-of-time.

A Neural-embedded Choice Model: TasteNet-MNL Modeling Taste Heterogeneity with Flexibility and Interpretability