傳統上，工廠在分配作業員至產線時通常是採隨機分配或刻板印象，如:男生被視為強壯因此被分配至組裝、搬運站。女生則被視為細心，因此會分配至外觀檢查站。然而，每個人都有他們天生的才能及不同的生長背景，這些因素都影響在特定站位下的表現，如果分配在不適合的站位時，極有可能使該作業員的離職率提高或產品良率下降，進而造成工廠成本上升。在此篇論文，我們設計了一套基於機器學習、大數據及統計方法的系統，並使用工廠的人力資源歷史資料，建立一個隨機生存森林的模型，進而自動尋找每個作業員最適合的站位，期望能降低離職率。經過工廠的AB測試，如果使用我們的分配系統，可以降低約3%的離職率，p-value為0.0014，此數字也達統計學上的顯著差異水準，證明我們所設計的系統是有效的。

The traditional operators’ allocation method in our factory to product line is random or by population stereotype, for instance, males be deemed to be more strength and should be placed at assembly line, and females should stay in cosmetic checking line since who are more circumspect. Nevertheless, individuals have their natural talent and growing background that would affect their performance if they were allocated in stations unsuitable to them, furthermore, cause attrition rate and cost growth. In this thesis, we design a system that using the knowledge of Big Data Analysis, Machine Learning and Statistics to create a Random Survival Forest model and the human resource data of our factory as training data that could automatically allocate every operator to their most appropriated station based on their individual features. After adopting this system into the factory with an A/B testing, the result shows that this scientific way can surely reduce the attrition rate from 12.7% to 9.7%, and the p-value is 0.0014 that reaches the significant level of Hypothesis Testing, which proofs that our experiment is effective and successful.

[1] Kayoko Murakami, “A solution of human resource allocation problem in a case of hotel management”, The 40th International Conference on Computers and Industrial Engineering, pp.1 - 6, 2010

[2] Lixin Zhou, “A Project Human Resource Allocation Method Based on Software Architecture and Social Network”, The 4th International Conference on Wireless Communications, Networking and Mobile Computing, pp.1-6, 2008

[3] Yi Zhang, “Research on human resource allocation optimization based on genetic algorithm from the perspective of two-way choice model”, 2010 International Conference on Educational and Information Technology (ICEIT), pp.V1-380 - V1-383, 2010

[4] Guozheng Zhang, "Customer Retention Based on BP ANN and Survival Analysis", 2007 International Conference on Wireless Communications, Networking and Mobile Computing, pp. 3406 - 3411, 2007

[5] Ganthan Narayana Samy, "A framework for integrated risk management process using survival analysis approach in information security", 2010 Sixth International Conference on Information Assurance and Security (IAS), pp. 185 - 190, 2010

[6] David Cox, "Regression Models and Life-Tables", Journal of the Royal Statistical Society, Series B 34, pp.187 - 220, 1972

[7] Leo Breiman, “Random forest”, Machine Learning, pp.5 - 32, 2001

[8] Mohammad S. Alam, "Random Forest Classification for Detecting Android Malware", 2013 IEEE International Conference on Green Computing and Communications and IEEE Internet of Things and IEEE Cyber, Physical and Social Computing, pp. 663 - 669, 2013

[8] Zhenping Yi, "Application of random forest to stellar spectral classification", 2010 3rd International Congress on Image and Signal Processing (CISP), pp.3219 - 3132, 2010

[10] Gajendra Jung Katuwal, "The predictive power of structural MRI in Autism diagnosis", 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp.4270 - 4273, 2015

[11] Ning Chang, “The Application of Neural Network to the Allocation of Enterprise Human Resources”, 2010 2nd International Conference on E-business and Information System Security, pp.22 - 23, 2010