階層貝氏聯合分析法中潛藏偏好係數之估計

Estimation of Latent Part-worth in Hierarchical Bayes Conjoint Analysis

林致誠(Chih-Cheng Lin)

指導教授 ： 任立中

國立臺灣大學/共同教育中心/統計碩士學位學程/碩士(2017年)

https://doi.org/10.6342/NTU201701086

摘要

聯合分析法(Conjoint analysis)能夠衡量顧客對於產品不同屬性的偏好，並能協助公司建立市場模型來預測市佔率、盈餘、甚至新產品進入市場後的預期獲利。但是公司可能需要花費大量的時間和金錢以收集到足夠的資料，因此在實務上該如何用較少的問卷資料還原同樣質量的顧客異質性是相當重要的課題。本研究提出結合複合屬性的修正模型以縮短問卷量，並嘗試利用二階段階層貝氏模型還原複合屬性背後的潛藏偏好，以協助我們做出更精確的預測以及對於顧客偏好更全面的了解。本研究比較了修正模型以及原始模型的表現，修正模型在樣本內預測的表現勝過原始模型，但修正模型在樣本外預測的表現不及原始模型。

關鍵字

聯合分析法 ； 階層貝氏模型 ； 潛藏偏好 ； 複合變數 ； 多變量模型

並列摘要

Conjoint analysis can measure consumers' preference (part-worths) for different features of a product, which can help the companies to create market models that estimate market share, revenue and even profitability of newly-designed products or services. However, in order to gather enough information, the process of data-collecting can be time-consuming and cost lots of money, so practitioners would be eager to find a way to recover the heterogeneity in the part-worths with shorter questionnaires. In our study, we proposed a modified model using complex attribute to shorten the questionnaires. Also, we tried to recover latent part-worths behind the complex attribute with the use of two-stage hierarchical Bayesian model, which may help us make better predictions and better understanding the part-worths of customers. Compared to the original model, the modified model performed better at in-sample prediction, however, modified model performed weaker than original model at out-sample prediction.

並列關鍵字

Conjoint Analysis ； Hierarchical Bayesian Model ； Latent Partworth ； Complex Variable ； Multivariate model

參考文獻

Adams, L. L. M. and Gale, D. (1982). Solving the quandary between questionnaire length and response rate in educational research. Research in Higher Education, 17(3):231– 240.

Allenby, G. M., Arora, N., and Ginter, J. L. (1995). Incorporating prior knowledge into the analysis of conjoint studies. Journal of Marketing Research, pages 152–162.

Allenby, G. M., Arora, N., and Ginter, J. L. (1998). On the heterogeneity of demand. Journal of Marketing Research, pages 384–389.

Andrews, R. L., Ansari, A., and Currim, I. S. (2002). Hierarchical bayes versus finite mixture conjoint analysis models: A comparison of fit, prediction, and partworth recovery. Journal of Marketing Research, 39(1):87–98.

Berdie, D. R. (1989). Reassessing the value of high response rates to mail surveys. Marketing Research, 1(3).