模糊理論應用在百貨公司賣場 ─業種組合與樓層配置分析

Application of Fuzzy Cluster Theory on Department Retail Analysis of Business Category Combination & Floor Distribution

鍾明德(Ming-Te Chung)

指導教授 ： 古思明 博士

元智大學/工程學院/工業工程與管理學系/碩士(1997年)

摘要

本研究嘗試利用在工程控制領域方面的模糊理論的概念，將之應用在百貨公司上。此舉可以說是一種創新與嘗試，而不同以往商學的統計手法。將相近似（Similarity）的業種匯聚，劃分成為一族群（Cluster ），可以使所區分的「業種組合」讓消費者容易辨識。求得分類後的族群（業種組合）內有同質特性（Homogeneity ）而不同族群（業種組合）之間適當的隔離（Well-separated）。使特定客層的消費者能到特定的區域去惠顧，一面可以增加購物的便利性，另一方面可以刺激消費。

本研究利用模糊聚類分析中的『分佈樹』的概念，了解哪些業種適合放在同一樓層；再利用電腦『資料結構』的概念，利用父節點與子節點的關係，以及利用顧客惠顧偏好指數的大小來一次各個節點的重排。經過『模糊聚類分析』的複雜處理建立的分佈樹，接著加入『業種惠顧偏好指數』的分佈樹，由不同『 -cut』與『業種惠顧偏好指數』下建立分佈樹結果，可以作為百貨公司實務界對於「不同業種」對於「不同樓層」的安排作為參考。

研究結果顯示：
1. -cut：可以得知各個業種之間『關連性』的大小，正如研究方法所提出的三個通則， 值較大的放在同一層， 值較小的放在較鄰近層，或較遠的樓層。
2.業種惠顧偏好指數：業種惠顧偏好指數大的對應至低樓層，而業種惠顧偏好指數小的對應至高樓層。

實證應用結論：各個『業種組合』的『阻隔性係數』差異很小（應說是太小），這和原先實務界的構想有很大出入（消費者（因樓層的不同），惠顧行為有一定的『阻隔性』）。故百貨公司的立體動線良好，阻隔性並不是想像那麼大。

而各個『業種組合』的『業種惠顧偏好指數』差異明顯，百貨公司因應改裝，增加想要引進的專櫃時，可以安排至適當的樓層，最重要的考量並不是參考樓層效用比率的概念（高低樓層專櫃租金收費差異2 ~ 4倍）來利用阻隔性係數。應考量各個『業種組合』的『業種惠顧偏好指數』，來找一個惠顧偏好程度高或知名度高的專櫃。

關鍵字

百貨公司 ； 模糊聚類分析 ； 業種組合 ； 樓層配置

並列摘要

This study applies the [ distribution tree] concept of fuzzy cluster analysis in understanding the ideal floor for individual businesses. Computer''s information structure is further applied, using the relation between main points and sub-points along with the index of consumer visit favour in re-arranging the point distribution . The distribution tree is then built through repetitive and complex processing of fuzzy cluster analysis which is then further incorporated with the distribution tree of [ index of business type visit preference]. The different [α- cut ] and [ index of business type visit preference] are used for establishing the results as reference of arranging the floor distribution of different business categories in department stores.

The results show that:
1. α-cut： yields the relation scale among businesses which conform to the three common principles as proposed by this study that businesses of greater αvalues should be placed in the same floor, and lowerα values should be distributed in the neighbouring floors or farther floors.
2. Index of buesiness type visit preference： Business of greater index values should be placed in lower floors, and smaller to the higher.

Empirical Finding： The difference [ separating coefficiency] among individual businesses is marginal ( too insignificant) which greatly deviates from the original field belief that visiting behaviour is affected by certain isolation due to floor of location. This proves that the routing of department store is well disigned; the separation effect is not as great as perceived.

Meanwhile the difference between [ business combination] and [index of business type visit preference ] is quite apparent. When department stores wish to introduce new counters due to strategy modification, appropriate floor distribution can be arranged. The most critical concern is not the floor efficiency ratio ( Difference of rental cost for higher and low floors can be as great as 2 to 4 times ) in using separating coefficiency. The various [ indexes of business type visit preference ] of [ business combination ] should be taken into account in seeking out a business operation of greater consumer favour or publicity.

並列關鍵字

Department Store ； Fuzzy Cluster ； Analysis ； Business Combination ； Floor Distribution

參考文獻

（15） Peter R. Dickson & James L. Ginier, " Market Segmentation, Product

Differentiation, and Marketing Strategy", Journal of Marketing, Vol.51（April 1987）, pp. 4-5.

（17） Kotler, P. （1989）, "From Mass Marketing to Mass Customization,"Planning

IL：Richard D. Irwin.

（20） Klir, G. J. and Folger, T. A. " fuzzy sets, uncertainty, and information",