Intra metropolitan transferability of mode choice models

Abstract

This article considers whether individual (mode) choice models are transferable to areas other than that for which they were estimated, evaluating the notion of transferability and testing it in two Sydney suburban areas. One of the main sources of model bias is found not to be simply due to different populations; dissimilarities between model coefficients in previous studies can be explained at least in part by differences in data collection procedures. The empirical study analyses two compatible data sets. Tests are described to evaluate the ability of binary- logit models to replicate from one sample the behaviour of a second sample of commuters. Transferred models do not perform as well on the second sample as models estimated unrestricted, because of specification errors. If transferability is judged by statistical criteria, the models estimated in this study are not transferable (A). (TRRL)

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 52 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.