ABSTRACT

Trip generation models are typically calibrated using zone-based aggregated data, despite some theoretical drawbacks pertaining to the use of aggregated data. This study demonstrates that models calibrated with disaggregated data not only have a better theoretical basis, but also forecast more accurate trip productions. In particular, optional trip purposes are focused on the calibration of trip production models, of which trips tend to be irregularly generated, rendering it difficult to build more reliable models. There are three models calibrated: an aggregate model with zone-level data, a disaggregate model with household-level data, and with household and person-level data using household survey data for two years. The prediction accuracy and stability of the models are compared and verified using statistical methods. It is discovered that calibrating a trip production model using household and person-level data jointly yields a better calibration method.

摘要

出行生成模型通常使用基于区域的聚合数据进行校准，尽管在使用聚合数据方面存在一些理论缺陷。本研究表明，用分解数据校准的模型不仅具有更好的理论基础，而且预测更准确的旅行产生。特别是，可选的出行目的侧重于出行生产模型的校准，其中出行往往是不规则地生成的，因此很难建立更可靠的模型。校准了三个模型：区域级数据的聚合模型、住户级数据的分解模型以及使用两年住户调查数据的住户和个人级数据。使用统计方法对模型的预测精度和稳定性进行比较和验证。