data mining: opportunities and challenges
Chapter XVI - Data Mining for Human Resource Information Systems
Data Mining: Opportunities and Challenges
by John Wang (ed) 
Idea Group Publishing 2003
Brought to you by Team-Fly


Only a few organizations have started data mining their HRIS. MERLIN, a data warehouse developed by the State of Mississippi, currently allows over 230 users to run and access aggregate reports on position vacancies, salary patterns, and other HR-related data (Roberts, 1999). In time, the State plans to use the data warehouse to help find trends to identify where it has retention problems, attrition rates and reasons, and average employee characteristics. HR Vision, an HR data warehouse product developed by SAS Institute, has helped Deere & Co. find a way to quickly make predictions and run reports from over 80,000 employee records (SAS Institute, 2001). For example, the company identified patterns in employee-benefit selections. Through this process, it was able to provide benefit options that were more attractive to workers, and, as a result, it was better able to manage the costs of the options provided.

The questions in Table 3 outline some areas of exploration for data mining an HRIS. As an organization comes to the realization that it has a vast amount of information stored that it isn't using, it will probably be able to identify many other questions to be explored. As mentioned earlier, it is important to formulate a question before jumping into data mining. Here we will look at forming questions in two specific areas identifying effective sources of recruiting employees and estimating the value of training. These examples are fairly simple, but provide the reader with an idea of the how data-mining techniques can be used to support human resource decisions.

Employee Recruitment Support

Organizations are consistently looking for more effective recruiting methods to minimize recruiting costs and also to find employees that are more likely to stay with the organization for an extended period of time. High employee turnover can be very costly to companies. The cost of turnover goes beyond the fee for running an advertisement in a newspaper. Turnover costs include all replacement costs such as staff time in interviewing, the cost of lost productivity while the position remains open, and the cost of training the new employee (Cascio, 1991). Yet, very few employers evaluate the success of their recruiting sources.

There are many sources from which organizations fill open positions. An organization may place an advertisement in a newspaper or on the Internet, hold a job open house, collect employee referrals, use a third-party recruiter, or many other options. Recruiters may use a variety of approaches in very large companies without ever communicating to each other the success (or lack of success) of a particular source. Data mining offers a possible solution to this problem. If an organization tracks the recruitment source of new hires, it can search for patterns in recruitment source relating to successful employees. To do this, the organization must first identify the measure to identify successful employees. An organization might, for example, define a successful employee as an employee who received a top performance appraisal rating.

At least two different data-mining methods could be used for this question. First, a regression model or Neural Network model might be applied. In this approach, a success score would be assigned to employees in the data sample. Then the possible predictor variables would be decided. One or both of the above model types could be run on the data to obtain a formula that predicts the success score of an employee with the given predictor variable values. Secondly, a classification type of approach could be used. In this modeling approach, two samples one of high performers and one of low performers are obtained. Then, a statistical group score is developed. The data for a new potential hire can then be processed to predict in which group he or she will be.

Employee Training Evaluation

An important step in the instructional design process is the evaluation of training. Most training evaluation focuses on evaluating specific training programs and the specific change that has or has not occurred as a result of that program (Sackett & Nelson, 1993). This type of evaluation is helpful in determining the content of future training programs by understanding the components of specific training programs that were effective or ineffective. This type of evaluation is essential to organizations, but some organizations may want to look at the value of training in a different context.

Employee training and development activities in an organization are of significant strategic importance, and they are also very costly to organizations (Tannenbaum &Woods, 1992). Considering the high cost of training in an organization, company leaders may be interested in looking at training from the "big picture" perspective. If a company can identify its overall training budget, it may want to know what kind of impact this total figure is having on the company. One approach to identifying this effect would be to look at the success of employees who have participated in company training programs. Are they progressing through the organization? Are they receiving promotions and moving up the career ladder?

This type of information is not readily available to organizations. While they may be able to evaluate the results of a specific training effort, it is often difficult to evaluate training as a whole. Data mining might offer a solution to this evaluation question. Many organizations store training-related data in their HRIS. Some of the data stored includes the content of training programs and records of which courses an employee has taken (Herren, 1989). By setting up a data-mining program to search for patterns of training activities related to advancement in the organization, a company might uncover data to support further training investments.

Brought to you by Team-Fly

Data Mining(c) Opportunities and Challenges
Data Mining: Opportunities and Challenges
ISBN: 1591400511
EAN: 2147483647
Year: 2003
Pages: 194
Authors: John Wang © 2008-2017.
If you may any questions please contact us: