A Study on Staff Turnover Prediction by Example of the Employees of an Information Product Distributor

• Main Authors: Cheng-Feng Wang, 王政豊
• Other Authors: Prof.Patrick S Chen
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/51198479122473333266

碩士 === 大同大學 === 資訊經營學系(所) === 103 === Talented professionals have always been an important cornerstone of business growth. Turnover will result in reduction of competitiveness， increase in recruitment and training costs. Therefore， how to avoid loss of talents is an important topic in research and an appreciable issue in practice. In view of the previous studies，most researchers only explore factors affecting turnover. This cannot prevent talents from quitting their job effectively. Some researches focus on the relationship between turnover and personal properties such as age， gender， aptitude， family， etc. This only gives the hindsight and cannot accurately forecast the turnover for a particular employee， to its best， only speculating a group of people with these characteristics. This study assumes that the external behaviors of an employee are a reflection of his intrinsic psychological complex. As a result，job burnout， frequent leaves，working overtime， and browsing job websites are factors to be analyzed. In addition to the results of the previous studies that form the basis of employee work behavior attributes， we add external behaviors such as overtime， attendance records， e-mails，Internet behaviors， etc.， to construct an employee turnover prediction system. Through literature review we learn the key factors that affect job termination of employees. Then， we select corresponding attributes in personnel database and transform the data to avoid privacy violation. The data are collected from a major IT product distributor. Study objects are the employees who quit their job in the year of 2013 1nd 2014. Their behavior of work overtime， attendance record， late for work， mail record，browsing history， etc. are examined in relation with the time of leave. The result shows that turnover is associated with frequency of late for work and website visits in correlation to type of work， performance， and year of employment. It is possible to predict job termination of employees through periodical and long-term observation. Suggestions for preventing of job turnover are also provided.