模糊推論應用於消費者決策法則之研究

An Inquiry into Consumer Decision Rules with Fuzzy Inference

陳亭羽(Ting-Yu Chen) ； 崔哲偉(Che-Wei Tsui)

《管理科學研究》 2007特刊 (2007/12) Pp. 11-34

https://doi.org/10.6873/MSR.200712/SP.0002

摘要

消費者即使面對相似的購買決策情況，但在不同問題解決類型之下，消費者所採取的決策法則仍可能有很大的差異。因此，本研究欲探討消費者對於特定產品在不同問題解決類型所採取的決策法則差異為何，及利用模糊推論以交叉驗證的方式預測在不同問題解決類型之下，以影響消費者採取決策法則之因素，其可能採取之決策法則之結果。本研究共回收有效問卷87份。實證資料顯示，廣泛問題解決類型(EPS)整體產品涉入分數為4.5；有限問題解決類型(LPS)整體產品涉入分數為3.9；例行問題解決類型(RPS)整體產品涉入分數為4.0；EPS與LPS整體知覺風險分數約為4.3，而RPS僅為3.2；EPS與LPS整體知覺風險分數約為5.0，而RPS僅為3.6；EPS下的品牌熟悉程度平均分數為3.8；LPS下的品牌熟悉程度平均分數為4.2；RPS下的品牌熟悉程度平均分數為3.6；EPS下追求最佳產品極大化的程度平均分數為4.6；LPS下追求最佳產品極大化的程度平均分數為4.3；RPS下追求最佳產品極大化的平均分數為3.6。本研究以模糊推論之十六種的不同模糊蘊含式預測消費者面對特定產品下，其可能採取決策法則為何。經模糊推論交叉驗證後，本研究發現S-蘊含式之有界加法與R-蘊含式之標準交集、代數乘積、有界差異預測結果表現較好。整體而言，R-蘊含式的整體預測結果表現為最好。

關鍵字

購買決策 ； 問題解決類型 ； 決策法則 ； 模糊推論 ； 模糊蘊合式

並列摘要

Understanding how buyers make their consumption decisions and adapting to consumer needs are absolute necessities for competitive survival in marketplaces. According to the amount of available information and the speed in making a decision, consumer decision making can be distinguished into three specific levels: extensive problem solving, limited problem solving, and routine problem solving. Consumers consider sets of product attributes by using different decision rules, depending on the complexity of the problem solving process and the importance of the choice to them. Therefore, this study attempts to differentiate decision rules based on the influences of antecedent determinants (such as involvement and perceived risk) under each type of problem solving processes. Fuzzy inferences with a generalized modus ponens procedure are applied to facilitate approximate reasoning of decision rules. Since fuzzy implications are fundamentals of rule reasoning, we discuss inference effects concerning sixteen basic types of fuzzy implications. A total of 90 questionnaires were sent out and 87 valid copies retrieved. From the empirical analysis, we found that desirable prediction results occur frequently in bounded sum operation of S-implications and standard fuzzy intersection, algebraic product, and bounded difference operations of R-implications. In conclusion, the inference results of decision rules using antecedent determinants indicate that R-implications have significantly effect on prediction accuracy under each level of problem solving processes.

並列關鍵字

consumption decision ； problem-solving level ； decision rule ； fuzzy inference ； fuzzy implication