Individualized Hybrid Models for Conjoint Analysis
提出并比较了四种混合模型，用于处理包含大量属性的联合分析研究，通过行业研究评估其内部交叉验证、市场份额估计和属性重要性聚类等性能。
Increasingly, conjoint analysts are being asked to design and analyze studies involving large numbers of attributes and/or attribute levels. Various types of approaches, including attribute bridging, Adaptive Conjoint Analysis, and hybrid models have been proposed to deal with the problem. This paper describes recent developments in hybrid modeling. Four hybrid models are described and compared in terms of their performance in an industry-based study entailing 15 product attributes. Comparisons are made in terms of internal cross-validation, market share estimates, attribute importances clustering, and its relationship to exogenous background variables. The proposed models are also compared to selected models from the transportation science literature. The authors emphasize the point that comparative model performance may strongly depend on the ways in which the models are to be used.