1-6 各群對企業收益表現比較 - 以改良式RFM模式結合資料探勘建立客戶分群並提升行銷效益

由附圖 5.1-5 可知，高價值客戶群中對的企業收益的表現遠大於其餘三群客戶的收益表現總和，由於此表乃是針對交易資料來加總，故所代表之意義為各群實際為企業的貢獻程度。由此表可知，高價值客戶所帶來的貢獻大於其餘三群貢獻的總和，而可推論只要針對高價值客戶群投入最大的心力，

即可相對貢獻於企業的收益當中。然而對其餘三群的客戶，可採取不同的客戶管理策略。

有了此四群的客戶資料，即可針對此四群的客戶採取不同的市場策略。

如高價值客戶即是企業長時間的獲利來源，企業必須盡全力來保持這些客戶

的滿意度與忠誠度。主力型客戶所扮演的角色是交易頻繁程度及交易次數可能不及高價值客戶所給企業帶來的利益，但可能在單一指標上，如交易金額有相當可觀的利益，企業也必須對此類型的客戶加以提升其客戶滿意度，以期望在往後能給企業帶來更多的利益。無價值客戶所代表的的是無論從哪一個指標來看，此類型的的客戶並無法帶給企業顯著的利益，故此類型在行銷策略上是採取放縱的策略，並不特意經營此類型各戶。對於成長型客戶而言，

此類型的客戶對企業在短期無法為企業帶來顯著的利益，但相對具有潛力能提供企業未來利潤，可以說是企業值得培養的對象，相對而此類型的客戶，

建議企業可以採取行有餘力才關注的可客戶群，可先針對高價值客戶及主力型客戶投入較大的心力與資源。

5.2. 行銷決策支援分析

分析企業客戶的類型，會發現有許多的客戶對公司來說是沒有利潤的，

這些客戶應該要逐步的捨棄，但是有幾類的客戶，我們會把它們保留下來︰

z 參考指標客戶 z 推薦型客戶 z 可學習的客戶

有些客戶可以提高我們的聲望，它們通常是深具影響力的大型客戶，這便是所謂的參考指標客戶，像是供貨給 Wal-Mart 或是 HP，雖然賺不到什麼錢，

但是其它的客戶聽到我們跟這些公司合作，會覺得我們公司是值得信賴的公司，因此可以帶來更多的客戶，進而帶來更多的利潤，這類型的客戶就像是我們的廣告招牌，不可以輕易割捨。某些客戶會幫我們介紹新的客戶，或是在其它場合推薦本公司，我們稱它們為推薦型客戶。和它們交易可能不賺錢，

但是卻可能是最有產值的不支薪業務。

某些客戶擁有了值得學習的營運機制，或是它們在創新上經常有嶄新的突破，跟它們做生意，就像是聘顧了一位免費經營顧問，從它們身上我們不止可以學習到科技的創新、生產程序上的創新，更重要的是可以學到管理方法上的創新，以及行銷服務上的創新，有了這樣的良師益友，可以強健公司的體質，讓我們在未來可以擁有更強的競爭力，當然不可以輕易捨棄。

6. 結論與建議

6.1. 研究結論

此客戶分群的模型中，應用了類神經網路的自組織特徵映射圖網路

(self-organizing map network,SOM)作為客戶分群的演算法，並結合 RFM 指標來評估客戶對企業的貢獻程度，並將分群結果進一步做客戶管理的決策資訊。

由以上的分析結果，可以推論此分群方法可以獲得相當良好的客戶分群結果，即使我們於資料前處理時所發現客戶 RFM 之指標分佈並非常態分佈，

也由此可驗證 SOM 方法所具有的資料分析強健性，並不被限制於資料的常態性。

另外我們只針對這些客戶分成無價值客戶、成長型客戶、主力型客戶、

高價值客戶四群，當然 SOM 的分群能力可將客戶分成多群，但以現有應用反而增加爾後分析的負擔，且分成四群對於決策者能做更有效的判斷。

6.2. 研究限制與建議

本文中所提出的客戶分群模型固然具有強健與效率性，但基於取樣資料的限制，在往後的研究中仍可朝以下方向繼續深究：

z RFM 分群指標的強化

在原始的 RFM 指標中，其三個指標並無權重的關係，亦即何者的指標在特定的產業當中有較重要的地位。舉例來說，以直銷業來說，每次交易

的金額可能差異均不大（因著產品的價格維持在一個水準），但此時分別客戶重要程度的應是購買頻率（Frequency）較其餘兩指標更為重要。

是否存在對於不同產業的特性，因而有不同的 RFM 權重的研究也相對重要。[2002, 陳彤生]在他所提的改良式 RFM 模型中利用決策樹歸納法加權 RFM 模式、吉尼指標法加權 RFM 模式來評估顧客效益，並結合 SOM 是一個未來的研究方向。

z SOM 分群模型的強化

SOM 分群演算法雖然具有強健性，且不用理會原始的資料分佈情形（如在本文中所舉的例子即為偏左的分佈），且輸出結果可有多樣性(XML、

SVG、pdf 等)，但若客戶的交易中相似性（Similiarity）過高，則易導致分群的結果中同一群的客戶數目過高（在此例中，為了區分出無價值客戶、成長型客戶、主力型客戶、高價值客戶等四群客戶分類，故指定 SOM 產生的群數為四，且此例的結果產生了高價值客戶的數目為 350 個、

主力型客戶共 5 個、無價值客戶共 42 個、成長型客戶共 66 個）。而在此分析結果中客戶的數目過多反而造成市場決策的困難（依此例而言，

企業反而投入大部分的精力與時間在高價值客戶上，而無法區別出就有影響力的 20％的客戶群）。因此如何在特定群中決定出適當的客戶數是一個研究課題，並期望對應於 Preto 所提的 80/20 法則，提供企業一個利用最少的資源投入於對企業最有幫助的客戶群。

7. 參考文獻

[1] Dittenbach, M., Rauber, A. & Merkl, D. (2002). Uncovering hierarchical structure in data using the growing hierarchical self-organizing map, Neurocomputing, 48(1-4), pp199-216.

[2] Everitt, B. S., Landau, S., & Leese, M. (2001). Cluster analysis.

4th ed., London: Edward Arnold.

[3] Halkidi, M., Batistakis, Y., & Vazirgiannis, M. (2001). On

clustering validation techniques. Journal of Intelligent Information Systems, 17(2/3), pp107-145.

[4] Jain, A. K. & Dubes, R. C. (1988). Algorithms for clustering data.

Prentice-Hall advanced reference series. Upper Saddle River, NJ:

Prentice-Hall.

[5] Jain, A. K., Murty, M. N., & Flynn P. J. (1999). Data clustering: a review. ACM Computing Surveys, 31(3), pp264-323.

[6] Robert Zeithammer( 2004 ), “Customer analysis:

Segmentation”

,BUS 37000: Marketing Strategy

[7] Hughes, Arthur M.（1994）, Strategic Database Marketing, Chicago：Probus Publishing.

[8] Stone, Bob（ 1995 ）, Successful Direct Marketing Methods, Lincolnwood, IL：NTC Business Books, pp37-59.

[9] Ron Kohavi, Rajesh Parekh (2004),”Visualizing RFM

Segmentation",

Proceedings of the 2004 SIAM International

Conference on Data Mining (SDM'2004)

, Orlando, FL, pp391-399.

[10] Vesanto, J. (2000). Using SOM in data mining, Licentiate’s thesis, Helsinki University of Technology, Finland.

[11] Vesanto J. & Alhoniemi, E. (2000). Clustering of the

self-organizing map. IEEE Transactions on Neural Networks, 11(3), pp586-600.

[12] Z.-P. Lo, M. Fujita and B. Bavarian (1991): Analysis of neighborhood interaction in Kohonen neural networks. 6th International Parallel Processing Symposium Proceedings, pp 247-249.

[13] 李章偉（2000），「資料庫行銷之顧客價值分析：以 3C 流通業為例」，國立臺灣大學國際企業學系研究所碩士論文。

[14] 謝依真（2000），「不同分群方法與不同資料來源之比較」，國立成功大學工業管理學系研究所碩士論文。

[15] 劉世琪（2002），「應用資料挖掘探討顧客價值：以汽車維修業為例」，朝陽科技大學工業工程與管理系研究所碩士論文。

[16] 連惟謙（2003），「應用資料分析技術進行顧客流失與顧客價值之研究」，中原大學資訊管理研究所碩士論文。

[17] 趙景明（2003），「應用資料探勘技術於顧客價值分析之研究」，

東吳大學資訊科學系碩士論文。

[18] 陳彤生(2002)，「運用改良 RFM 提升行銷效益的實證研究」，朝陽科技大學主辦第七屆人工智慧與應用研討會。

[19] 林慧晶（1997），「資料庫行銷之客戶價值分析與行銷策略應用」，

國立臺灣大學國際企業學系研究所碩士論文。

[20] 楊東昌(2004)，「自組織映射圖神經網路改善模式與分群應用之回顧研究」，華梵大學工業管理學系碩士論文。

8. 附錄

附錄一 GNU GENERAL PUBLIC LICENSE 通用公共許可證 Version 2, June 1991

51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.

Preamble

The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Lesser General Public License instead.) You can apply it to your programs, too.

When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.

To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain

responsibilities for you if you distribute copies of the software, or if you modify it.

For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.

We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.

Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by

someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.

Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.

The precise terms and conditions for copying, distribution and modification follow.

TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION

0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law:

that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as

"you".

Activities other than copying, distribution and modification are not covered by this

License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does.

1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program.

You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.

2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions:

a) You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change.

b) You must cause any work that you distribute or publish, that in whole or in part

在文檔中以改良式RFM模式結合資料探勘建立客戶分群並提升行銷效益 (頁 62-76)