Teacher Performance Evaluation Analysis Using K-Means Clustering Algorithm and Random Forest Classification

Abstract

Teacher performance assessment is a primary parameter in determining the quality of educational institutions. Evaluation systems in many elementary schools still rely on descriptive qualitative approaches. Abundant school administrative data often remain as unprocessed archival records without further analytical utilization. This condition results in school management decision-making that lacks a strong empirical foundation. This study applies data mining technology to transform administrative data into strategic information. The research focuses on SD Negeri 12 Padang Besi and involves all active teaching staff during the current academic year. The research dataset is entirely derived from internal school records. This study excludes the use of questionnaire instruments, and in-depth interview methods are not employed in the data collection process. The analysis is strictly limited to administrative aspects, without including assessments of in-class pedagogical competence. The technical implementation utilizes the K-Means Clustering algorithm to automatically identify patterns in teacher performance grouping. This process is followed by the application of the Random Forest algorithm to measure classification accuracy based on the available administrative features. The combination of these methods produces a performance mapping that is free from human subjectivity. The analytical results provide clear performance labels for each individual teacher. This study contributes to the development of a data-driven digital evaluation model. School management can use the outputs of this research as a basis for reward allocation or targeted professional development programs. This approach ensures transparency in human resource governance within the educational environment.