Toward explainable generative foundations for large-scale transportation modelling

This thesis builds tools that make transport models more robust and reliable. It creates three methods: SAA, TreeCSP, and ActVAE that generate the people, households, and daily activities used in travel simulations. Each is designed to be transparent so users can see how decisions are made. SAA builds realistic populations, TreeCSP creates realistic household relationships, and the ActVAE produces daily activity patterns that reflect real behaviour. Together, these methods help future transport models better represent diverse communities while remaining explainable, adaptable, and reliable for policy and planning.