Identifikasi Faktor Literasi Digital Siswa Pasca Pelatihan dengan Algoritma Random Forest

Abstract

Digital literacy is a foundational competence for junior high school students as learning increasingly relies on digital platforms; however, empirical evidence identifying which measurable factors most strongly drive post-training improvement remains limited. This study aims to determine key predictors of digital literacy gains after structured training and to develop a predictive model that classifies improvement into three levels (low, moderate, high). Data were collected from 200 junior high school students who participated in a structured program in digital marketing and graphic design, comprising pre-test and post-test scores, participation indicators, learning motivation, and frequency of digital tool use. After data cleaning, transformation, and feature encoding, a Random Forest classifier was trained to model improvement categories. Model performance was assessed using an 80:20 train–test split and stratified five-fold cross-validation, reporting accuracy, precision, recall, F1-score, and confusion matrix analysis. The model achieved 78% accuracy and exhibited its strongest and most stable performance in the high-improvement category, while minority categories showed reduced sensitivity, suggesting the influence of class imbalance.