Dissecting university employee attendance log: A case study

Abstract

Attendance is one of the many ways organizations use to evaluate the performance of their employees. In this study, we present the analysis of university employee attendance log to understand employee behavior before developing an attendance information system as well as to gain insight on policy refinement. Process mining technique is used to aid the analysis. From the analysis, we have identified some potential problems that may be encountered when processing the data and formulated several solutions to mitigate them. We also take a closer look on how the employees behave and whether they conform to the university policy. Finally, we have proposed modified prediction methods to address the problem of missed clocks. Our proposed method managed to lower the prediction overestimation rate to around 16% while keeping prediction accuracy within acceptable range.