隨著經濟的發展和資料科學技術的進步，企業在市場上的競爭更加激烈，在數據多樣化、移動網絡技術與設備普及的今天，可否借助深度學習(Deep learning)技術讓多層次傳銷企業高層管理可以從運營資料數據中掌握住影響分銷商升級成功的特徵因子，進而調整在分銷商管理運營時之策略規劃。

本研究提出灰度表徵學習多層次傳銷分銷商升級模型，利用企業資源規劃（Enterprise Resource Planning, ERP）資料，蒐集涉及分銷商晉級相關之歷史數據，採集成學習（Ensemble learning ）概念，將灰色關聯度分析(Grey Relational Analysis, GRA)與隨機森林(Random Forest, RF)之重要因子算法兩者結果整合成GRA RF Ensemble Learning(GARF)特徵篩選規則，進行特徵篩選。篩選出在營運、分銷商與分銷產品的各層重要表徵值再導入貝葉斯網路結構學習法進行建構。最終，該模型可導入領域專家經驗值或貝葉斯推論方法來推論出分銷商升級結果，並可透過可視化模型來瞭解影響分銷商成功升級各層面的重要營運特徵因子及關聯度，進而提供給多層次傳銷企業的高層運營決策者，讓他們能夠瞭解對分銷商升級影響的營運因子並調整分銷商服務及運營決策。

With the development of the economy and the advancements in data science technology, competition among enterprises in the market has intensified. In today's era of data diversity and widespread adoption of mobile network technology and devices, can deep learning techniques be leveraged to enable senior management in multi-level marketing (MLM) companies to effectively capture the key factors that influence distributor upgrade success from operational data? Subsequently, can strategies and plans for distributor management and operations be adjusted accordingly?
This study proposes a gray-scale representation learning model for multi-level MLM distributor upgrades. By utilizing Enterprise Resource Planning (ERP) data and collecting historical data related to distributor promotions, the study incorporates the concept of ensemble learning and integrates the results of the gray correlation degree and the important factor algorithm of random forest into a feature selection rule known as GRA RF Ensemble Learning (GARF). This rule is applied to perform feature selection. The significant representation values at different levels of operations, distributors, and distribution products are then selected and used in the construction of a Bayesian network structure using the Bayesian network structure learning method.
Ultimately, this model can incorporate domain expert knowledge or employ Bayesian inference methods to infer distributor upgrade outcomes. Moreover, it provides a visualized model to understand the important operational feature factors and their correlations that significantly influence the success of distributor upgrades at various levels. This information is valuable for senior-level decision-makers in multi-level marketing companies, enabling them to gain insights into the operational factors affecting distributor upgrades and adjust distributor services and operational decisions accordingly.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
目錄 vi
圖次 ix
表次 x
第一章 緒論 1
第一節 研究背景 1
第二節 研究動機 2
第三節 研究目的 3
第二章 文獻探討 4
第一節 傳銷與傳銷業績 4
第二節 多層次傳銷與組織 5
第三節 灰色系統理論與灰色關聯度分析 6
第四節 隨機森林(RF) 7
第五節 整合學習 8
第六節 貝葉斯網路 9
第三章 研究方法與步驟 11
第一節 個案研究之資料蒐集 11
第二節 資料不平衡處理 12
第三節 灰色關聯度分析及隨機森林重要因子 14
1.特徵選擇（Feature selection） 14
2.隨機森林法的特徵選擇方法 14
3.灰色關聯度分析法 15
4.GARF特徵因子篩選法：(GRA RF Ensemble Learning, GARF): 18
第四節 貝葉斯網路結構學習法 19
第五節 二元分類器模型效能指標評估 23
1.二元混淆矩陣(Confusion matrix): 23
2.接收器操作者性質曲線(ROC curve ) 和 曲線下面積(AUC Curve): 24
3. F1 分數(F one score): 24
4. Kappa(Cohen’s kappa)： 25
第六節 研究步驟與流程 26
第四章 實驗結果與討論分析 27
第一節 資料清理與資料不平衡處理 27
第二節 表徵因子篩選 32
第三節 學習貝葉斯網路結構之分銷商升級模型 35
第四節 分銷商升級模型結果評估 39
第五節 分銷商升級模型討論分析 43
第五章 研究結論與建議 46
第一節 在多層次傳銷產業的分銷商升級模型 46
1.分銷商管理最佳化： 46
2.對分銷商培訓與支援： 46
3.服務決策參考與可視化： 46
4.發展與創新： 46
5.方法整合： 47
第二節 本研究結果建議 47
1.不平衡資料問題: 47
2.貝葉斯網路最佳化: 47
3.考慮更多的影響因素： 47
4.擴大研究樣本： 48
5.分銷商升級制度： 48
參考文獻 49
附錄 55

[1]公平交易委員會，”中華民國110年多層傳傳銷事業發展狀況調查結果報告”
[2]吳漢雄、鄧聚龍、溫坤禮，灰色分析入門，台北：高立圖書公司，1996年
[3]陳得發, & 鄭旭棠. (1999). 多層次傳銷組織網的經營管理之研究. 中華管理評論, 一月. 2卷1期，1-20頁.
[4]覃怡輝. (1998). 傳銷實務制度的探討. 第四屆直銷學術研討會論文集, 第 35-60頁。台北市：中華直銷管理學會
[5]謝亞靜(2012) .直銷產業競爭優勢與經營策略之研究—以中國大陸安利為例(碩士論文）。國立交通大學企業管理碩士學程，臺北市
[6]鍾祥鳳 (2006). 每週及每月固定聚會和加捷科技多層次傳銷組織穩定成長的關連性研究(碩士論文)。國立交通大學，臺北市
[7]Biggart, N. W. (1989). Charismatic capitalism: Direct selling organizations in America. University of Chicago Press.
[8]Bouckaert, R. R. (1995). Bayesian belief networks: from construction to inference (Doctoral dissertation).
[9]Breiman, L. (2001). Random forests. Machine learning, 45(1), 5-32.
[10]Calle, M. L., & Urrea, V. (2011). Stability of Random Forest importance measures. Briefings in bioinformatics, 12(1), 86-89.
[11]Chang, K. C., & Yeh, M. F. (2005). Grey relational analysis based approach for data clustering. IEE Proceedings-Vision, Image and Signal Processing, 152(2), 165-172 .
[12]Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002). SMOTE: synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321-357.
[13]Chen, M. F., & Tzeng, G. H. (2004). Combining grey relation and TOPSIS concepts for selecting an expatriate host country. Mathematical and computer modelling, 40(13), 1473-1490.
[14]Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and psychological measurement, 20(1), 37-46.
[15]Coughlan, A. T., & Grayson, K. (1998). Network marketing organizations: Compensation plans, retail network growth, and profitability. International Journal of Research in Marketing, 15(5), 401-426.
[16]Cowell, R. G., Dawid, P., Lauritzen, S. L., & Spiegelhalter, D. J. (2007). Probabilistic networks and expert systems: Exact computational methods for Bayesian networks. Springer Science & Business Media.
[17]Cortes, C., & Mohri, M. (2003). AUC optimization vs. error rate minimization. Advances in neural information processing systems, 16.
[18]Cox, I. J., Miller, M. L., Minka, T. P., Papathomas, T. V., & Yianilos, P. N. (2000). The Bayesian image retrieval system, PicHunter: theory, implementation, and psychophysical experiments. IEEE transactions on image processing, 9(1), 20-37.
[19]Duan, L. & Xu, L.D. (2012) ‘Business Intelligence for Enterprise Systems: A survey’, IEEE Transactions on Industrial Informatics, 8(3), pp. 679–687.
[20]Fang, G., Guo, Y., Huang, X., Rutten, M., & Yuan, Y. (2018). Combining grey relational analysis and a bayesian model averaging method to derive monthly optimal operating rules for a hydropower reservoir. Water, 10(8), 1099.
[21]Fawcett, T. (2006). An introduction to ROC analysis. Pattern recognition letters, 27(8), 861-874.
[22]Friedman, N., Geiger, D., & Goldszmidt, M. (1997). Bayesian network classifiers. Machine learning, 29(2), 131-163.
[23]Guyon, I., & Elisseeff, A. (2003). An introduction to variable and feature selection. Journal of machine learning research, 3(Mar), 1157-1182.
[24]Good, W. S., & Hassay, D. N. (2015). Multilevel marketing plans: A new channel of distribution. In Proceedings of the 1995 World Marketing Congress (pp. 294-300). Springer, Cham.
[25]Gregor, B., & Wadlewski, A. A. (2013). Multi-level marketing as a business model. Marketing instytucji naukowych i badawczych, (1 (7)), 2-19.
[26]Han, H., Guo, X., & Yu, H. (2016, August). Variable selection using mean decrease accuracy and mean decrease gini based on random forest. In 2016 7th ieee international conference on software engineering and service science (icsess) (pp. 219-224). IEEE.
[27]Heckerman, D., Geiger, D., & Chickering, D. M. (1995). Learning Bayesian networks: The combination of knowledge and statistical data. Machine learning, 20(3), 197-243.
[28]Hosseini, S. (2021). A decision support system based on machined learned Bayesian network for predicting successful direct sales marketing. Journal of Management Analytics, 8(2), 295-315
[29]Japkowicz, N., & Stephen, S. (2002). The class imbalance problem: A systematic study. Intelligent data analysis, 6(5), 429-449.
[30]Kim, E. (2017) Essays on business networks in the multi-level marketing industry, UNIVERSITY OF MICHIGAN LIBRARY. Available at: https://deepblue.lib.umich.edu/bitstream/handle/2027.42/138716/eunsoo_1.pdf
[31]Kung, C. Y., & Wen, K. L. (2007). Applying grey relational analysis and grey decision-making to evaluate the relationship between company attributes and its financial performance—a case study of venture capital enterprises in Taiwan. Decision support systems, 43(3), 842-852.
[32]Lee, K. F., Lau, T. C., & Loi, K. Y. (2016). Driving distributors’ satisfaction in multilevel marketing (MLM) companies. International Journal of Academic Research in Business and Social Sciences, 6(2), 105-122.
[33]Liu, S., & Lin, Y. (2010). Introduction to grey systems theory. In Grey systems (pp. 1-18). Springer, Berlin, Heidelberg.
[34]Longadge, R., & Dongre, S. (2013). Class imbalance problem in data mining review. arXiv preprint arXiv:1305.1707
[35]Ma, S. C., & Shi, H. B. (2004, August). Tree-augmented naive Bayes ensembles. In Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat. No. 04EX826) (Vol. 3, pp. 1497-1502). IEEE.
[36]Menardi, G. & Torelli, N. (2012) ‘Training and assessing classification rules with Imbalanced Data’, Data Mining and Knowledge Discovery, 28(1), pp. 92–122. doi:10.1007/s10618-012-0295-5.
[37]Nat, P. J. V., & Keep, W. W. (2002). Marketing fraud: An approach for differentiating multilevel marketing from pyramid schemes. Journal of Public Policy & Marketing, 21(1), 139-151.
[38]Neapolitan, R. E. (2004). Learning bayesian networks (Vol. 38). Upper Saddle River: Pearson Prentice Hall.
[39]Nikovski, D. (2000). Constructing Bayesian networks for medical diagnosis from incomplete and partially correct statistics. IEEE Transactions on Knowledge and Data Engineering, 12(4), 509-516.
[40]Sahinbas, K. (2022, April). Employee Promotion Prediction by using Machine Learning Algorithms for Imbalanced Dataset. In 2022 2nd International Conference on Computing and Machine Intelligence (ICMI) (pp. 1-5). IEEE.
[41]Saito, T., & Rehmsmeier, M. (2015). The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets. PLoS ONE, 10(3), e0118432[66] Saito, T., & Rehmsmeier, M. (2015). The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets. PLoS ONE, 10(3), e0118432.
[42]Schwarz, G. (1978). Estimating the dimension of a model. The annals of statistics, 461-464.
[43]Seijo-Pardo, B., Bolón-Canedo, V., Porto-Díaz, I., & Alonso-Betanzos, A. (2015, June). Ensemble feature selection for rankings of features. In Advances in Computational Intelligence: 13th International Work-Conference on Artificial Neural Networks, IWANN 2015, Palma de Mallorca, Spain, June 10-12, 2015. Proceedings, Part II (pp. 29-42). Cham: Springer International Publishing.
[44]Shen, Q. & Daly, R. (2007) Aberystwyth University methods to accelerate the learning of bayesian ..., core.ac.uk. Available at: https://core.ac.uk/download/pdf/288843056.pdf
[45]Song, Q., & Shepperd, M. (2011). Predicting software project effort: A grey relational analysis based method. Expert Systems with Applications, 38(6), 7302-7316.
[46]Spirtes P, Glymour C, Scheines R (1993) Causation, Prediction and Search, Lecture Notes in Statistics, vol 81. Springer Verlag.
[47]Sreekumar, P. (2007). A study of Multi-Level Marketing (MLM), As a potential tool for socio-economic development).
[48]Tsai, C. H., Chang, C. L., & Chen, L. (2003). Applying grey relational analysis to the vendor evaluation model. International Journal of The Computer, The Internet and Management, 11(3), 45-53.
[49]Tuv, E., Borisov, A., Runger, G., & Torkkola, K. (2009). Feature selection with ensembles, artificial variables, and redundancy elimination. The Journal of Machine Learning Research, 10, 1341-1366
[50]Tzeng, G. H., & Tasur, S. H. (1994). The multiple criteria evaluation of grey relation model. The Journal of grey system, 6(2), 87-108.
[51]Uusitalo, L. (2007). Advantages and challenges of Bayesian networks in environmental modelling. Ecological modelling, 203(3-4), 312-318.
[52]Vaughan, B., & Wang, Q. (2008). Classification based on tree-structured allocation rules. Journal of Experimental Education, 76(3), 315–340.
[53]Vrieze, S. I. (2012). Model selection and psychological theory: a discussion of the differences between the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). Psychological methods, 17(2), 228.
[54]Watthayu, W., & Peng, Y. (2004, August). A Bayesian network based framework for multi-criteria decision making. In Proceedings of the 17th international conference on multiple criteria decision analysis.
[55]Wotruba, T. R., & Tyagi, P. K. (1991). Met expectations and turnover in direct selling. Journal of Marketing, 55(3), 24-35.
[56]Yang, C. C., & Chen, B. S. (2006). Supplier selection using combined analytical hierarchy process and grey relational analysis. Journal of Manufacturing Technology Management.
[57]Yang, Z., & Peterson, R. T. (2004). Customer perceived value, satisfaction, and loyalty: The role of switching costs. Psychology & marketing, 21(10), 799-822.
[58]Yu, J., Smith, V. A., Wang, P. P., Hartemink, A. J., & Jarvis, E. D. (2004). Advances to Bayesian network inference for generating causal networks from observational biological data. Bioinformatics, 20(18), 3594-3603.
[59]Zhang, Z. H., Wang, X. M., Dang, J. W., & Min, Y. (2014). Bayesian network structure learning method based on expert knowledge fusion. Computer Engineering and Applications, 50(2), 1-4.
[60]Zhu, R., Guo, Y., & Xue, J. H. (2020). Adjusting the imbalance ratio by the dimensionality of imbalanced data. Pattern Recognition Letters, 133, 217-223.