資料探勘技術如何降低組織人力流動支出？

How to use Data Mining Techniques to Reduce the Organization of Human Resource Mobility Expenses?

廖珮妏(Pei-Wen Liao) ； 陳玫華(Mei-Hua Chen) ； 于俊傑(Chin-Cheh Yi)

《服務業管理評論》 12期 (2014/12) Pp. 1-20

摘要

本文以個案需求為出發點，配合個案公司積極推廣資料探勘等預測模型來挖掘人事流程改善效能的方法，本研究順勢善加利用個案所具備的優勢，擁有健全的人事資料庫，來發現並改善人事缺口問題，包括挑選影響留任因素，以及尋找出留任率最佳數值點，有效降低人力養成成本。鑑於此，本文以資料探勘技術的倒傳遞類神經網路為分析方法，進行留任預測，串連個案公司既有的人事動態系統（HRS）及甄選招募（e-RS）系統，將這兩大類次級資料庫作為資料分析來源。經研究結果發現，影響留任的重要變數有雇用別、年齡、證照狀況、行為、情感、認知等六因素，而最佳留任數值點為0.797值，誤判率為1.9%，損失不到2%的人力。研究結果可供個案公司利用HRS以及e-RS作為新人招募參考依據。

關鍵字

次級資料庫 ； 留任 ； 類神經網路

並列摘要

In this research for the case study demand as the starting point, with the active promotion forecast model such as data mining to improve the performance of the human resource process. In this research make good use of conform advantage for case study which a sound personnel database to find the problem and improve the human resource gap and issues, to make the selection of retention factors, to find the best value point of retention and to predict the retention rates effectively. This research is based on the data mining technology of back-propagation neural network analysis methods to predict retention. The two database of HRS and e-RS system are to combine with a source of database analysis. We found that there are six factors affect retention: hire category, age, license status, behavior, emotion, cognition, and best value point of retention is 0.797; an error rate is 1.9% that does not less than 2% manpower. This result can provide the basis for recruiting reference through HRS and e-RS systems.

並列關鍵字

Secondary data ； Retention ； Artificial neural network