近幾年，工程建案急速增加，人才流動非常頻繁，營造廠商徵才面試持續不斷，倘只要有些疏忽或是流程時間掌控不當，很容易就錯失優秀人才錄取之時機，加上等待其錄取人才回應或是來公司報到時間拉長，甚至不易掌控是否要繼續應徵其他才，付出之時間成本甚鉅。本研究依照分類方式彙整收集多家北部營造業之面試數據資料，以最多12層次對每一分類做劃分，並運用四種最受歡迎之分類演算工具進行預測，其中包含了多層層感知器(MLP)、支援向量機(SVM)、隨機森林(RF)、自組織映射機(TS-SOM)，並且採用了五種交叉驗證方式來比較結果與分析，研究結果得知，以支援向量機(SVM)最為至準確，可達到93%預測面試人員錄取、非錄取與婉拒錄取之準確度，未來在招募徵才時，可運用營造公司自有之歷史面試數據，並搭配此計算機之運算方式輔助錄取審核結果之參考，可以準確判斷出錄取人才報到之機率性，減少企業等待中時間浪費。 ;The rapid increase in construction projects and frequent workforce turnover pose challenges for construction firms in continuous recruitment interviews. Any oversight or improper control of the process timeline can easily lead to missing the opportunity to recruit excellent talent. This study categorically compiles and collects interview data from multiple construction companies in the northern region, totaling 560 entries spanning five years (2018-2023). The study considers 14 influencing factors and utilizes four commonly used classification algorithms for prediction, including Multilayer Perceptron (MLP), Support Vector Machine (SVM), Random Forest (RF), and Self-Organizing Map (TS-SOM). Five cross-validation methods are employed to compare and analyze the results. The findings reveal that the Support Vector Machine is the most accurate, achieving a prediction accuracy of 93% for interviewees′ acceptance, rejection, and deferment. In the future, accurate assessment of the probability of recruited talent reporting can reduce the time wasted by companies during the recruitment process.
