Abstract
Machine learning enables organizations to leverage data strategically to improve employee performance, promote continuous improvement, and better fulfill the mission. Opportunities for leveraging machine learning within organizations exist throughout the employee lifecycle but should be pursued with a clear understanding of the strengths and limitations of the methodology. This chapter will review traditional performance management processes, introduce machine learning as a methodology, highlight how machine learning methods could be used in new performance assessment models, and note future research directions to improve the use of machine learning within organizations.
TopOne of the most common approaches to employee performance assessment is an annual planning and review model. Unsurprisingly, this approach follows from a common organizational financial approach in which the annual budget drives much of the decision-making over the subsequent year. As Hope and Fraser (2003) describe in Beyond Budgeting: How Managers Can Break Free from the Annual Performance Trap:
For most participants, the budgeting process is an annual ritual that is deeply embedded in the corporate calendar…. It typically begins at least four months prior to the year to which it relates…it starts with a mission statement that sets out some of the aims of the business. This is followed by a group strategic plan that sets the direction and high-level goals of the firm. These form the framework for a budgeting process that grinds its way through countless meetings at which points are traded as targets are negotiated and resources agreed upon. (p. 3)
They note that these targets then form the basis of “budget packs” that are distributed to various departments and units which subsequently define how success will be measured in the coming year. With such metrics in mind, departments can then incorporate such priorities and direction in determining employee priorities and ultimately how they would contribute to the organizational success.
The process for employees often follows analogously to the aforementioned budgetary one. Similar to how the budget process first involves an evaluation of how the organization did compared to the fiscal year plan, employee prior performance plans or goals are unearthed and they’re typically tasked to document specific achievements, deliverables, and activities as they relate to the plan. Supervisors then combine those materials with their own internal assessments, to determine how employees performed in year passed. Then, either individually or collaboratively, new goals are established for the upcoming year and documented in a manner to be revisited in the next annum. This too follows the budget planning cycle in that the planning is intended to flow from the organizational mission, down to the departmental level, and ultimately to the individual employee, thus providing a framework to ensure that every employee is proverbially rowing in the same direction and contributing to the overall effectiveness of the organization.
In concept, it would appear to be an elegant and constructive means of not only evaluating employee performance for promotions, raises, and constructive placement. Ideally it would also provide a regular means of reinforcing the inherent value of employees to the success of the institution and provide structure to determine how to reward, support, and equip employees in the year to come to make an even greater impact. Such activities should increase employee effectiveness and satisfaction, while also strengthening the human capital contribution to the mission and make the organization stronger and more capable of fulfilling its mission. As Pulakos (2004) noted in a review of performance management systems, “In fact, if developed and implemented properly, performance management systems drive employees to engage in behaviors and achieve results that facilitate meeting organizational objectives” (p. 5).
Key Terms in this Chapter
Training Dataset: The portion of the input dataset used to train the machine learning model.
Testing Dataset: The portion of the input dataset used to test the effectiveness of the trained model against previously unseen data.
Supervised Learning: A type of machine learning which uses a labeled dataset, such that the algorithm attempts to match the output labels based on input data.
Algorithmic Bias: The phenomenon when a machine learning algorithm produces biased output based on the biased input data.
Training a Model: The process of processing input data using a machine learning algorithm in order to generate a model that minimizes error.
Machine Learning: Use of computer algorithms to automatically discover patterns in data based on mathematical processes.
Explainable AI (XAI): An emerging approach where one artificial intelligence system evaluates the underlying code of another artificial intelligence or machine learning model and generates a human-interpretable explanation of the outcomes.
Decision Tree: A machine learning algorithm that cascades data through a series of sequential decisions to determine classification or regression outcomes.
Unsupervised Learning: A type of machine learning in which the algorithm attempts to find patterns in the data without predefined outputs.
Performance Management: A process of planning, tracking, and evaluating goals and performance in order to ensure that employees are effectively contributing to the organizational mission.