Evidence on Transferability of Trip-Generation Models

Abstract

Full and partial transfer results of trip-production models among intercity, intracity, and intraregional locations are reported in this paper. Nineteen models were transferred 208 times among the locations. The variety of transfer conditions and number of transfers permitted inferences to be drawn on the conditions influencing transferability. It was observed that the quality of local data seriously affects the ability to discern transferability, and may account for the variable results observed in the literature. It was also observed that model quality and income affect transferability. For all transfers conducted in this study, the transferred models explained an average of 57% of the variation in trip-making behavior explained by locally estimated models. When locally estimated constants were used to replace transferred constants in the model, the transfer models explained, on average, 87% of the variation in trip-making behavior explained by locally estimated models. It was concluded that the adjustment of transfer-model parameters with local information (partial transfer) has the potential to improve transferability to such an extent that the need to predict transferability becomes redundant.

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 — 24 citations exist upstream.

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