本研究目的在於構建運具分群與旅客區隔的整合模式,以探討臺灣西部走廊的城際運輸旅運行為。在運輸文獻中,多項羅吉特模式的IIA特性能使模式校估較為便利,並簡化誤差項假設。然而,此特性忽略旅運者在選擇過程中,會有運具方案相似問題,也未考慮到個體的隨機偏好異質,進而可能導致有用資訊遺失與校估結果不一致。 因此,有別於單獨進行運具分群與旅客區隔之個別模式,本研究首先透過巢式羅吉特模式針對方案間是否具有相似性進行探討;再利用潛在區隔模式與混合羅吉特模式考慮個體間異質性的存在,經由程序性校估方式以構建考量運具分群與旅運者區隔之整合模式,以瞭解各運具方案間的屬性相似程度與旅運者對於運具的偏好異質。 本研究實證結果發現,在城際運輸旅運選擇行為中,確實存在著方案相似性與個體異質性的問題。然而,研究結果亦證明透過運具分群與旅客區隔的程序校估方式分析旅運者選擇行為,能有效解決方案相似性與個體異質性的問題,且有效提升模式的解釋能力。
This study aims to construct an integrated mode choice model which considers mode cluster and traveler segmentation for intercity traveling behavior in Taiwan western corridor. With regard to those literatures for discrete choice model, multinomial logit model has great advantage over parameters estimation due to its i.i.d. assumption on error terms. However, this assumption neither conducts alternative similarity nor considers individual heterogeneity, and would lead to the defects of missing variable and biased estimator. Therefore, this study first uses nested logit model to investigate the alternative similarity and then applies latent class model and mixed logit model to conduct individual heterogeneity. This two-stage process with mode cluster and traveler segmentation could effectively investigate alternative similarity among modes and preference heterogeneity across travelers. The empirical results have shown that alternative similarity and individual heterogeneity indeed exist in the intercity traveling behavior in Taiwan western corridor. The proposed models which considering mode cluster and traveler segmentation can resolve the problems of alternative similarity and individual heterogeneity and also effectively increase the explanatory power of mode choice models.