<![CDATA[Abstract: The use of information technology devices can support students’ learning by providing greater access to information and more flexible academic activities. However, their use still requires appropriate management, particularly in relation to control, discipline, dependency, and social interaction. This study aims to identify patterns of information technology use, predict cluster membership, describe the characteristics of each cluster, and analyze trends in students’ learning outcomes. The study involved 272 students from Pesantren Diniyyah Puteri Padang Panjang, with data collected through questionnaires and documentation of learning outcomes. The analysis employed a Hybrid Machine Learning approach consisting of K-Means clustering based on nine variables of information technology use, Backpropagation to predict cluster membership, and the C4.5 Decision Tree algorithm to generate rules describing cluster characteristics. The K-Means analysis produced three clusters: light monitoring with 155 students, intensive assistance with 16 students, and targeted guidance with 101 students. The percentages of students with good learning outcomes in these clusters were 86.45%, 93.75%, and 79.21%, respectively. The Backpropagation model achieved an accuracy of 94.55%, while the Decision Tree generated rules that described the characteristics of each cluster. The findings indicate that good learning outcomes are not always accompanied by favorable patterns of information technology use across all dimensions. These results can serve as a basis for providing appropriate guidance on information technology use according to students’ characteristics. Keywords: information technology, learning outcomes, K-Means, Backpropagation, DT. Abstrak: Penggunaan perangkat teknologi informasi dapat mendukung pembelajaran santri melalui akses informasi dan kegiatan akademik yang lebih fleksibel, namun tetap memerlukan pengelolaan terkait kontrol, disiplin, ketergantungan, dan interaksi sosial. Penelitian ini bertujuan mengidentifikasi pola penggunaan teknologi informasi, memprediksi keanggotaan klaster, menjelaskan karakteristik klaster, serta menganalisis kecenderungan hasil belajar santri. Data penelitian melibatkan 272 santri Pesantren Diniyyah Puteri Padang Panjang yang diperoleh melalui kuesioner dan dokumentasi hasil belajar. Analisis menggunakan pendekatan Hybrid Machine Learning, yaitu K-Means berdasarkan sembilan variabel penggunaan teknologi informasi, Backpropagation untuk memprediksi keanggotaan klaster, dan Decision Tree C4.5 untuk membentuk aturan karakteristik klaster. Hasil K-Means membentuk tiga klaster, yaitu pemantauan ringan sebanyak 155 santri, pendampingan intensif 16 santri, dan pembinaan terarah 101 santri, dengan persentase hasil belajar baik masing-masing 86,45%, 93,75%, dan 79,21%. Model Backpropagation mencapai akurasi 94,55%, sedangkan Decision Tree menghasilkan aturan karakteristik setiap klaster. Temuan menunjukkan bahwa hasil belajar yang baik tidak selalu disertai pola penggunaan teknologi informasi yang baik pada seluruh dimensi. Hasil penelitian dapat menjadi dasar pendampingan penggunaan teknologi informasi sesuai karakteristik santri. Kata kunci: teknologi informasi, hasil belajar, K-Means, Backpropagation, DT.]]>
Copyrights © 2026
