以模糊規則為基礎之資料探勘分析消費者於產品因子之偏好

Abstract

藉由消費者對產品與產品各因子的偏好評估，本文提出一以模糊規則為基礎之資料探勘方法，可以由所收集之問卷中找出簡化的模糊規則，並找出定義於產品因子上的模糊語意值組合及其相對應之消費者效用。本文的主要目的在於發展一有用之決策分析工具，使決策者可應用前述之資訊以制定適當之行銷或經營策略，而在做法上係發展以基因演算法為基礎之資料探勘技術，決定一些使用者難以設定之參數，以自所收集之資料中找出前鑑部與後鑑部分別為語意值組合與該組合相對應之消費者效用之簡化模糊規則。此外，產品的意義並不限定為實物，而是廣泛的解釋為一切可被評估之對象。本文以速食店服務品質評估之問卷資料為實驗例，由資料中找出具較大或較小消費者效用之語意值組合並進行分析，以說明所提出之方法在支援決策分析上的實用性與可行性。

This paper proposes a fuzzy data mining method that can find utilities of various combinations of linguistic values in product factors. The main aim of the proposed fuzzy rule-based method is to develop a useful decision making tool by discovering simplified fuzzy if-then rules, whose consequent part is a real number, from numerical data by using the proposed data mining technique based on genetic algorithm to automatically determine several pre-specified parameters. Then, the antecedence and consequence of a fuzzy rule are referred to as a combination of linguistic values and the corresponding utility of the antecedent part, respectively. To demonstrate the usefulness of the proposed methods for supporting decision makers to make appropriate marketing strategies, we use computer simulations on questionnaire data for evaluating service quality of fast food stores to analyze possible combinations linguistic values that can acquire larger or smaller utilities.