<![CDATA[Employee performance assessment is a strategic function in human resource management. However, many companies, including PT. Wijaya Manggala Premier Lestari, still conduct manual evaluations based on limited indicators such as total working days. This study aims to classify employee performance using the K-Nearest Neighbors algorithm based on attendance data from the Mekari Talenta platform. The primary dataset consisted of one hundred twenty two employee records from March 2026, with attendance data from the three preceding months (December 2025 to February 2026) used to verify the consistency of individual attendance patterns. The methodology followed the Cross-Industry Standard Process for Data Mining framework. Eight numerical features were extracted from attendance codes: total small late, total big late, total paid leave, total absence, total unpaid leave, total working day deductions, total working days, and attendance percentage. The K-Nearest Neighbors model was evaluated using five fold stratified cross-validation with K values of 3, 5, and 7. The model with K = 5 achieved the highest performance among the tested K values, with an average accuracy of 89%, precision of 91%, recall of 88%, and F1 score of 89% across all five folds. The confusion matrix confirmed that the model effectively distinguishes between good and poor performers. This research provides a practical, automated classification framework that transforms raw attendance logs into objective performance insights using the Java programming language.]]>
