基於價值距離衡量之動態多準則決策方法及其應用

A dynamic multi-attributes decision making method based on value distance and its applications

尹亮(Liang Yin)

指導教授 ： 徐煥智

共同指導教授 ： 鄭啟斌(Chi-Bin Cheng)

淡江大學/商管學院/管理科學學系博士班/博士(2019年)

https://doi.org/10.6846/TKU.2019.00364

摘要

為取得最令人滿意之決策結果，決策者的風險態度而不僅僅是待選項目的效用價值應被納入考量中。本研究使用Ｓ形曲線價值函數來取代傳統多屬性決策方法中的期望效用函數以反映決策者的風險趨避和風險追求行為。在此基礎上，為了進一步減輕使用者在標量參考點上遇到的困難，本研究使用價值函數和權重加總方法來定義每個待選項目相對於極端可行解的心理價值距離以衡量它們的總體展望價值。該方法的效能在比較分析和敏感度分析中得到了驗證，證明其能夠幫助減少多屬性決策中的常見的問題例如排序顛倒問題．之後，該方法被擴展到群體決策領域，將多位決策者的偏好加總後得出公正的解決方案。實驗證明了該方法是適當且穩定的。最後，為了處理現實社會中存在的動態多階段決策場景．本研究將此方法發展為動態多階段決策方法並應用在一個挑選海量資料服務建構商的實際標案過程中，前一輪決策結果被以回饋機制帶入到下一輪的決策過程中。使用者接受了最終結果並認為該決策過程是易用且有幫助的。

關鍵字

多屬性決策 ； 展望理論 ； 動態決策 ； 群體決策

並列摘要

To achieve the most satisfying decision results, not only the utility value of the alternatives but also the risk attitudes of the decision makers need to be considered. In this proposed model, the s-shape value function is adopted to replace the expected utility function that is often used in traditional MADM methods to reflect the risk-averse and risk-seeking behavior of decision makers. On top of that, to further reduce the user burden of identifying the reference points, the psychological value distance is defined to measure the overall prospect values of each alternative reference to extreme feasible solutions using the value function and the additive weighting method. Performance of the proposed algorithms is comparatively analyzed and sensitivity analysis is conducted, to prove that this mechanism can help reduce issues like rank reversal. After that, the method is extended to a group decision setting, and the preferences of multiple decision makers are aggregated to produce a fair result. The experiments show that it is an appropriate and robust MADM method. Finally, considering the real world dynamic decision making scenario, the model is further developed to be dynamic (can handle more than one rounds of decision making, as defined in another research of a dynamic multiple-criteria decision making framework) (Campanella and Ribeiro, 2011), and then was applied in a big data service provider selection bidding case and the results from previous decision making process were carried to the following round using a feedback mechanism. The users accepted the final results and were satisfied with the easy and helpful decision making process.

並列關鍵字

MADM ； Prospect theory ； Group decision ； Dynamic decision making

參考文獻

References

1. Abdellaoui, M. (2000). Parameter-free elicitation of utilities and probability weighting functions. Management Science, 46, 1497–1512.

2. Abdellaoui, M., Bleichrodt, H., & Paraschiv, C. (2007). Measuring loss aversion under prospect theory: A parameter-free approach. Management Science, 53(10), 1659–1674.

3. Alanazi, H. O., Abdullah, A. H., & Larbani, M. (2013). Dynamic weighted sum multi-criteria decision making: Mathematical model. International Journal of Mathematics and Statistics Invention, 1(2), 16.

4. Baky, I.A. and Abo-Sinna, M.A. 2013. TOPSIS for bi-Level MODM problems. Applied Mathematical Modelling, 37: 1004–1015.