小型電商是否有能力把交易資料進行機械學習，得出經營的洞見? 小型電商要在與大型電商的競爭中存活，必須靠著比大型電商更靈活的策略或差異化來存活。就資源的角度來看，以往小型電商所擁有的資源往往不如大型電商般充足。近年來機械學習領域的快速發展，已經大幅下降了利用機械學習來改善決策品質的進入門檻，讓我們不禁疑惑這些小型電商如今是否能夠利用現有資源將經營資料進行機械學習，獲得以往只有大型電商才有能力做的資料分析和預測。

本研究試著在沒有專業級計算機硬體和不寫程式碼的狀態下，使用案例公司中實際的印表機耗材交易資料集，驗證小型電商是否也能夠利用Microsoft Azure Machine learning studio雲端計算技術搭配機械學習來預測重要資訊未來的走勢。選定價格彈性為預測標的，配合商品品牌、商品價格區間、銷售平台、季節性因素等四項因數。依照雲端分析技術的步驟: 取得資料、預處理、定義功能、選擇及套用學習演算法、評價模型、部署模型，一步一步分析出價格需求彈性的未來變化。並透過同時使用數種不同演算法來進行預測結果，最後使用最佳演算法提升決策樹迴歸分析得出未來的價格彈性預測值。利用Microsoft的POWER BI軟體做視覺化分析後，將結果建立桌上端及移動裝置的互動式視覺化儀表板。

經過驗證，小型電商的確可以在無專業電腦和不需要寫程式的狀態下，完成利用雲端技術建立預測模型，經由分析得出預測值，進而將整套系統佈署至雲端。這種方式大幅降低了機械學習的成本，也促成了機械學習入門門檻的持續降低。未來隨著資料蒐集系統和演算法的不斷完整化，整個機器學習的重心將由資料的蒐集和分析推展至預測結果的解釋。而負責解釋最終分析結果的產業專家或解釋各種人類行為的專家將扮演極其重要的角色。

This article is about to prove that a tiny e-commerce company can analyze online trading data and give business management valuable insight. To compete with giant e-commerce company, tiny company needs to fit fast-changing modern business world by flexible strategy and functional differentiation. In aspect to limited resource, general tiny company didn’t have enough resource to fight with their strong competitors. Since obviously progress in machine learning (ML) field recently, entry barriers of ML is not as high as before. We wonder that if tiny company can gain good quality predictive results by analyzing key managerial numbers or trading data in ML way as their richer competitors.

Thus, we design a tour of experience completely step-by-step ML procedures with real trading datasets which provided by a small hot copy (printer) consumables supplier in Taiwan. In order to simulate lack of resource situation, we won’t use high-end computers, write programming codes or demonstrate complex algorithms during whole process. In the middle of process, we will run Microsoft Azure machine learning studio cloud computing technology to help us simplify a part of ML flow and cut the time of calculating.

In this case, we choose price elasticity as predict target, analyzing the relationship with brand, price, platform and seasonal factors. Follow steps of Azure ML, from data receiving, preprocess, feature define, apply algorithm, evaluate model to develop model. In process, we also compared several algorithms simultaneously to evaluated best target prediction quality. After obtaining prediction result, we use Microsoft POWER BI to visualize result and establish inter-reactive dashboard for both desktop and mobile device.

As a result, we proved even tiny company can use cloud ML technology to build predictive model by itself without limits. By cloud computing ML method, it cuts ML cost into acceptable level for tiny company or individuals. As data collecting and Algorithm continued steadily grow into well-developed technology, next dilemma of ML will move to result explanation. In a future, industrial and humanities-related expert will play an important role in ML field.

國立中山大學研究生學位論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
第一章 緒論 1
第一節 研究背景及動機 1
第二節 研究目的 2
第三節 研究流程 3
第二章 文獻探討 4
第一節 電商業現況發展 4
第二節 機器學習的意義、分析應用與管理意涵 9
第三節 Microsoft Azure Machine Learning Studio介紹 17
第四節 價格彈性 22
第五節 商業智慧BI與POWER BI軟體 22
第三章 研究方法 26
第一節 研究架構 26
第二節 資料蒐集與個案簡介 27
第三節 敘述性統計 28
第四節 分析方法 37
第四章 研究結果與討論 42
第一節 Azure ML模型建立參數 42
第二節 預測結果 45
第三節 商業智慧與視覺化分析 49
第五章 研究結論 53
第一節 研究結論與管理意涵 53
第二節 未來研究建議 54
參考文獻 55

中文文獻:
方博亮、林祖嘉 (2017)。管理經濟學 (第五版)。智勝文化出版社。
克里夫頓 (2016)。透視數據下的商機：運用Google Analytics發掘商業洞見。 (譯者，林威利)。天下雜誌出版社。
科特勒，凱勒 (2017)。Marketing Management 行銷管理 (第 15e 版)。(譯者，徐世同、楊景博)。新北市: 華泰文化。
財團法人資訊工業策進會（民105）。我國B2C網路商店經營現況研析報告。105
張郁芝 (2014) 。 我國發展電子商務之現況與未來趨勢-以第三方支付為例。 國家發展委員會經濟發展處經濟期刊，14，109-129。
許碧芬、林彥甫、杜文弘、賴振安與陳柏元 (2003) 。大者恆大、贏者通吃的易遊網-以組織文化的觀點探討。第三屆觀光休閒暨餐旅產業永續經營學術研討會，高雄，國立高雄餐旅學院。
陳傑豪 (2015) 。 大數據玩行銷。30觀點出版社。
經濟部商業司(2016)。105年度電子商務雲端創新應用與基礎環境建置計畫（編號：III1051-P404-012）。臺北市：經濟部商業司。
謝邦昌 (2011)。雲端運算的應用-資料採礦和商業智慧。科學發展期刊，463，32-39。
謝邦昌、鄭宇庭、宋龍華、陳妙華 (2018)。大數據分析Excel Power BI全方位應用。碁峰資訊出版社。
簡禎富、許嘉裕 (2014)。資料挖礦與大數據分析。前程文化出版社。
簡錦漢、劉玉晳 (2017)。巨量資料在經濟學研究的應用及挑戰。人文與社會科學簡訊，9:1，6-12。

英文文獻:
Abelló, A., Darmont, J., Etcheverry, L., Golfarelli, M., Mazón López, J. N., Naumann, F., ... & Vossen, G. (2013). Fusion cubes: towards self-service business intelligence. DOI: 10.4018/jdwm.2013040104
Barga, R., Fontama, V., Tok, W. H., & Cabrera-Cordon, L. (2015). Predictive analytics with Microsoft Azure machine learning. Apress.
Barnes, J. (2015). Microsoft Azure Essentials Azure Machine Learning. Microsoft Press.
Bongers, F. Brischke, F. Bulcke, J.Militz, H. (2016). Evaluation of field test data. Paper prepared for the 47th IRG Annual Meeting.5
Borenstein, S. (2009). To what electricity price do consumers respond? Residential demand elasticity under increasing-block pricing. Preliminary Draft April, 30, 95.
Davenport, T. H., & Ronanki, R. (2018). Artificial Intelligence for the Real World. Harvard Business Review, 96(1), 108-116.
Fernández-Delgado, M., Cernadas, E., Barro, S., & Amorim, D. (2014). Do we need hundreds of classifiers to solve real world classification problems. J. Mach. Learn. Res, 15(1), 3133-3181.
Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189-1232.
Huang, Y. K., Kuo, Y. W., & Xu, S. W. (2009). Applying Importance-performance Analysis to Evaluate Logistics Service Quality for Online Shopping among Retailing Delivery. International Journal of Electronic Business Management, 7(2).
Luhn, H. P. (1958). A business intelligence system. IBM Journal of Research and Development, 2(4), 314-319.
McAfee, A., & Brynjolfsson, E. (2017). Machine, platform, crowd: Harnessing our digital future. WW Norton & Company.
McAfee, R. P., & Velde, V. (2008). Dynamic pricing with constant demand elasticity. Production and Operations Management, 17(4), 432-438.
Murray, S. (2017). Interactive Data Visualization for the Web: An Introduction to Designing with. " O'Reilly Media, Inc.".
RobertP.Christian. (2014). Machine Learning, a Probabilistic Perspective. CHANCE, 27, 62-63.
Samuel, A. L. (1959). Some studies in machine learning using the game of checkers. IBM Journal of research and development, 3(3), 210-229.
Suthaharan, S. (2014). Big data classification: Problems and challenges in network intrusion prediction with machine learning. ACM SIGMETRICS Performance Evaluation Review, 41(4), 70-73.
Watson, H. J., & Wixom, B. H. (2007). The current state of business intelligence. Computer, 40(9), 96-99.
網路資料:
微軟公司 (2015) 。 Power BI 導引式學習。取自網址https://docs.microsoft.com/zh-tw/power-bi/guided-learning/gettingstarted?tutorial-step=1#step-1
微軟公司(2017)。什麼是 Azure Machine Learning Studio。取自網址 https://docs.microsoft.com/zh-tw/azure/machine-learning/studio/what-is-ml-studio
微軟開發者論壇 (2015)。從 Azure Machine Learning 開始進入機器學習。2015年1月26日，取自網址https://blogs.msdn.microsoft.com/msdntaiwan/2015/01/24/azure-machine-learning/
經理人雜誌 (2018) 。「大數據」到底與我有什麼關聯。2016年4月6日，取自網址https://www.managertoday.com.tw/articles/view/52306。
資策會產業情報研究所 (2018)。網購大調查系列。取自網址 https://mic.iii.org.tw/IndustryObservations_PressRelease02.aspx?sqno=488

INTERNATIONAL DATA CENTER. (2017). Worldwide Hardcopy Peripherals Market Increases in Q3 2017. Retrieved from https://www.idc.com/getdoc.jsp?containerId=prUS43259217
Mitchell-Guthrie, P. (2018). Looking backwards, looking forwards: SAS, data mining, and machine learning. Retrieved from https://blogs.sas.com/content/subconsciousmusings/2014/08/22/looking-backwards-looking-forwards-sas-data-mining-and-machine-learning/
Raschka,S. (2018). Model evaluation, model selection, and algorithm selection in machine learning. Retrieved from sebastianraschka: https://sebastianraschka.com/pdf/manuscripts/model-eval.pdf
RD, P. (2018). [ #Office365 ] Power BI 2.0, part 1: the big picture (updated with GA infos). Retrieved from https://yos365.wordpress.com/2015/06/25/office365-power-bi-v2-part-1-the-big-picture/
Srivastava, T. (2018). What Is The Difference Between Machine Learning & Statistical Modeling. Retrieved from https://www.analyticsvidhya.com/blog/2015/07/difference-machine-learning-statistical-modeling/