決策樹、羅吉斯迴歸與類神經網路預測員工績效之比較研究

NCUIR > School of Management at National Central University > Graduate Institute of Human Resource Management > Electronic Thesis & Dissertation > Item 987654321/73672

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/73672

Title:	決策樹、羅吉斯迴歸與類神經網路預測員工績效之比較研究
Authors:	陳冠吟;Chen, Guan-Yin
Contributors:	人力資源管理研究所
Keywords:	資料探勘;人力資源管理;決策樹;羅吉斯迴歸;類神經網路;data exploration;human resource management;decision tree;logistic regression;neural network
Date:	2017-06-13
Issue Date:	2017-10-27 12:09:22 (UTC+8)
Publisher:	國立中央大學
Abstract:	人力資源領域中將資料探勘的分類技術應用於各方面並未相當常見。本研究將運用個案公司所提供人事資料庫之資料作為研究樣本，經由資料的蒐集及彙整過後，將資料進行分割，主要拆分為訓練樣本及驗證樣本兩部分，並以決策樹、羅吉斯迴歸、類神經網路等三種資料探勘技術建構員工績效高低預後模型。　　結果顯示，以決策樹及類神經網路預測員工績效高低情形的模型為最佳，兩種模型準確度都有90%，ROC曲線下面積AUC值分別為0.907和0.914，表示決策樹及類神經網路模型不管在準確度或是AUC值之兩種模型好壞評估標準上都屬於擁有優良預測能力的模型。 ;Using classification data-mining algorithm in predicting employee performance is rare. This study uses personnel data as research sample. After data cleaning and compiling process, data is divided into training dataset and the verification dataset. Then, this study uses three data mining technologies including decision tree, logistic regression and neural network to build employee performance prediction model by using training dataset. 　　The results show that the model of decision tree and neural network are the best in predicting employee performance by using verification dataset. Two accuracy of two model is 90%. Moreover, AUC is 0.907 and 0.914. It indicates that decision tree and neural network model have better prediction ability than logistic regression.
Appears in Collections:	[Graduate Institute of Human Resource Management ] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	228	View/Open

社群 sharing

Loading...