Employee Performance Classification Using Optimized Random Forest with Max Depth and N-Estimators at PT Bringin Gigantara

Abstract

Employee performance assessment is a crucial component of human resource management because it supports strategic decisions such as contract renewal, incentives, and competency development. This study proposes an optimized Random Forest model for employee performance classification at PT Bringin Gigantara by tuning the max_depth and n_estimators hyperparameters. The dataset consists of monthly performance evaluation records from 100 contract employees during January–December 2024, including ten performance variables. During model development, hyperparameter optimization was conducted exclusively on the training set using Stratified 5-Fold Cross-Validation to evaluate parameter combinations and select the most stable and accurate configuration. The optimization results indicate that the best-performing configuration was n_estimators = 50 and max_depth = 10, achieving a mean cross-validation accuracy of 98%. The optimized model achieved 98.75% accuracy on the training set and 95% accuracy on the test set, with corresponding precision, recall, and F1-score values demonstrating good classification performance. The findings indicate that the optimized Random Forest model can provide a practical, objective, and data-driven decision-support tool for human resource personnel in evaluating employee performance and supporting contract renewal decisions. Nevertheless, the model is intended to support, rather than replace, managerial judgment.