Penerapan Model 10-Fold Cross-Validation dalam Memprediksi Strategi Belajar Siswa SMA Berdasarkan Aspek Self-Regulated Learning (Manajemen Sumber Daya)

Abstract

Selecting the appropriate learning strategy is crucial for high school students; however, educators often struggle to identify strategies that align with each individual's Self-Regulated Learning (SRL) capabilities. This study aims to predict high school students' learning strategies based on the SRL aspect of resource management by applying the 10-fold cross-validation (C4.5) algorithm using the Knowledge Discovery in Databases (KDD) method. The study utilized 537 valid data points collected from students of SMA Negeri 1 Singaparna. A predictive model was constructed using 23 predictor attributes and evaluated through 10-fold cross-validation to assess its performance and reliability. The results indicate relatively low model performance, with an accuracy of 50.9% and an F1-score of 49.9%. This performance is attributed primarily to class imbalance and feature overlap among the learning strategy categories. The analysis identified the attribute "understanding improves when studying with peers" as the root node, with an Information Gain value of 0.042. These findings suggest that the resource management aspect of SRL is insufficient to serve as a sole predictor of students' learning strategies. Future research is recommended to incorporate other dimensions of SRL to improve predictive accuracy and provide a more comprehensive understanding of students' learning strategy preferences.