DEVELOPMENT OF SOFTWARE AGENT-BASED PERFORMANCE EVALUATION MODEL

Abstract

Agent-based performance evaluation systems help an organisation's predetermined uniform appraisal score weights but have some deficiencies, like difficulty in simulation tests, amongst others. The purpose of this study is to design computer-based systems for the effective performance evaluation of employees in an organization. The regression model was applied, and the results reveal that employee appraisal elements "must be objectives" and "tools for development" have a statistically significant positive influence on performance evaluation. This implies that an increase in "must-be objectives" and "tools for development" will correspondingly lead to an improvement in performance evaluation. Besides, the Artificial Neural Network (ANN) outperformed the other computer-based systems as it had an error bias of negligible value of 0.001. Meanwhile, in comparing the empirical data analysis results to the simulation, it is observed that the estimated parameter values of the regression model align with the findings, hence enhancing the precision of the model in achieving a steady state of the Ordinary Differential Equation (ODE). Additionally, the computer-based models are a good fit for the data, and the introduction of the three models shows better precision for performance evaluation in an organization. Therefore, the organisation should equip itself with the required knowledge base in computer-based systems to connect the real-life situation of predicting effective and unbiased performance evaluation.