近年來，公司愈來愈重視商品在保固期內被顧客退貨的責任歸屬，從客訴資料中分析多個變數資訊，以釐清異常品的問題。本研究旨在提高電池異常客訴的判定率，進而提升顧客滿意度與公司盈利。
　　由於個案公司的現行檢視方法與專業人員的經驗有關，十分容易造成誤判，目前已知電池膨脹第一階段可辨識率約為55%，第二階段可辨識率為60至70%，因此，本研究建議使用馬氏田口系統（Mahalanobis-Taguchi-system, MTS）、羅吉斯迴歸（logistic regression）與類神經網路（artificial neural network, ANN）來建立判定模式。此外，本研究經由一實際案例來說明三種方法的有效性，在馬氏田口系統、羅吉斯迴歸與類神經網路之訓練集準確度分別達到86.97%、86.55%及91.18%；測試集準確度依序為83.33%、79.41%和86.27%，此結果顯示本研究所建立之模型可有效辨別產品品質。最後，本研究討論三法的優缺點，並比較各項績效指標，以及特徵篩選後的重要特徵變數。

In recent years, companies have paid more and more attention to the attribution of responsibility for products returned by customers during the warranty period, and analyzed multiple variables from customer complaint data to clarify the problem of abnormal products. This research aims to improve the judgment rate of abnormal battery complaints, and thereby increases customer satisfaction and company profitability.
Since the current inspection method of the case company is related to the experience of professionals, and can cause misjudgments easily. At present, it is known that the identification rate of the first stage of battery expansion is about 55%, and the identification rate of the second stage is 60% to 70%. Therefore, this study proposes to apply Mahalanobis-Taguchi-system (MTS), logistic regression and artificial neural network (ANN) to establish a product judgment model. In addition, this study proves the effectiveness of the three methods through an empirical case. The training accuracy of the MTS, logistic regression and neural network reaches 86.97%, 86.55% and 91.18% respectively, and the testing accuracy is 83.33%, 79.41% and 86.27% in sequence. The results show that the established models can effectively identify product quality. Finally, this study discusses the advantages and disadvantages of the three methods, and compares various performance indicators, as well as important characteristic variables after feature selection.

Agresti, A. (2006). An Introduction to Categorical Data Analysis, Second Edition. John Wiley & Sons, Inc. https://doi.org/10.1002/0470114754
Berry, M. J., & Linoff, G. S. (2004). Data Mining Techniques: for Marketing, Sales, and Customer Relationship Management. John Wiley & Sons.
Chen, L.-F., Chen , Y.-C., & Su, C.-T. (2010). Mahalanobis-Taguchi System for the Prediction of Pressure Ulcers Development in Surgical Patients. 24th European Conference on Operational Research, Lisbon, Portugal.
Ching, R. K., Cheng, L., Ni, S.-F., & Chen, J. (2007). Applying Data Classification Techniques for Churn Prediction in Retailing. ICEB 2007 Proceedings (Taipei, Taiwan).
Cook, D. F., Ragsdale, C. T., & Major, R. (2000). Combining a neural network with a genetic algorithm for process parameter optimization. Engineering Applications of Artificial Intelligence, 13(4), 391-396.
Dubbs, D. (2001). Many (unhappy) returns. Operations and Fulfillment. 9. https://multichannelmerchant.com/operations/many-unhappy-returns/
Fausett, L., & Fausett, L. V. (1994). Fundamentals of Neural Networks: Architectures, Algorithms, and Applications. Prentice-Hall. https://books.google.com.tw/books?id=ONylQgAACAAJ
Fayyad, U. (1997). Data mining and knowledge discovery in databases: implications for scientific databases. Proceedings. Ninth International Conference on Scientific and Statistical Database Management (Cat. No. 97TB100150).
Fayyad, U., Piatetsky-Shapiro, G., & Smyth, P. (1996). From data mining to knowledge discovery in databases. AI Magazine, 17(3), 37-37.
Fayyad, U. M., Piatetsky-Shapiro, G., & Smyth, P. (1996). Knowledge Discovery and Data Mining: Towards a Unifying Framework. KDD.
García, S., Luengo, J., & Herrera, F. (2015). Data Preprocessing in Data Mining (Vol. 72). Springer. https://doi.org/10.1007/978-3-319-10247-4
Han, J., Pei, J., & Kamber, M. (2011). Data Mining: Concepts and Techniques. Elsevier.
McClelland, J. L., & Rumelhart, D. E. (1989). Parallel Distributed Processing: Explorations in the Microstructure of Cognition: Foundations (Vol. 1). MIT Press. https://ieeexplore.ieee.org/servlet/opac?bknumber=6276825
Milovanovic, M., & Peric, N. (2019). Binary Logistic Regression as a Method of Predicting Customer Dissatisfaction in Resolving Complaints: The Case of Bosnia and Herzegovina, Serbia and Former Yugoslav Republic of Macedonia. Calitatea, 20(173), 27-31.
Mizushima, K., Jones, P., Wiseman, P., & Goodenough, J. B. (1980). LixCoO2 (0< x<-1): A new cathode material for batteries of high energy density. Materials Research Bulletin, 15(6), 783-789.
Pratama, A. D., Al Zami, F., Supriyanto, C., Soeleman, M., & Renaldy, R. (2021). Classification Using Backpropagation Neural Network on Tweet Emotion Intensity. 2021 International Seminar on Application for Technology of Information and Communication (iSemantic),
Smith, G. (2018). Step away from stepwise. Journal of Big Data, 5(1), 1-12.
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. The Journal of Machine Learning Research, 15(1), 1929-1958.
Su, C.-T. (2013). Quality Engineering: Off-Line Methods and Applications. CRC press.
Su, C.-T., & Hsiao, Y.-H. (2007). An evaluation of the robustness of MTS for imbalanced data. IEEE Transactions on Knowledge and Data Engineering, 19(10), 1321-1332.
Su, C.-T., Hsu, H.-H., & Tsai, C.-H. (2002). Knowledge mining from trained neural networks. Journal of Computer Information Systems, 42(4), 61-70.
Taguchi, G., Chowdhury, S., & Wu, Y. (2001). The Mahalanobis-Taguchi System. McGraw-Hill. https://books.google.com.tw/books?id=5AOuyyccV8kC
Taguchi, G., & Jugulum, R. (2002). The Mahalanobis-Taguchi strategy: A Pattern Technology System. John Wiley & Sons.
Wongsasuluk, P., Chotpantarat, S., Siriwong, W., & Robson, M. (2018). Using hair and fingernails in binary logistic regression for bio-monitoring of heavy metals/metalloid in groundwater in intensively agricultural areas, Thailand. Environmental Research, 162, 106-118.
Woodall, W. H., Koudelik, R., Tsui, K.-L., Kim, S. B., Stoumbos, Z. G., & Carvounis, C. P. (2003). A review and analysis of the Mahalanobis—Taguchi system. Technometrics, 45(1), 1-15.
林大貴(2017)。Tensorflow+Keras深度學習人工智慧實務應用，博碩文化。
張云濤、龔玲(2007)。資料探勘原理與技術，五南圖書出版股份有限公司。
梁浩斌、杜建华、郝鑫、杨世治、涂然、张认成(2021)。锂电池膨胀形成机制研究现状，储能科学与技术，10(2)，647。
陳明達(2021)。應用馬氏距離於客訴產品之判定-個案研究，清華大學工業工程與工程管理學系學位論文。