Prediksi lama studi mahasiswa menjadi aspek penting dalam perencanaan akademik, evaluasi kinerja, serta identifikasi potensi keterlambatan kelulusan. Penelitian ini mengusulkan pendekatan hibrida dengan mengintegrasikan metode Rough Set dan Case-Based Reasoning untuk meningkatkan akurasi prediksi kelas lama studi mahasiswa. Metode Rough Set digunakan untuk mengekstraksi aturan klasifikasi berbasis kombinasi atribut IPK, status bekerja, dan status beasiswa, serta menghitung probabilitas kelas pada boundary region. Di sisi lain, metode CBR dimanfaatkan untuk menghitung similarity antar kasus berdasarkan kemiripan atribut, termasuk jumlah SKS yang dinormalisasi. Hasil prediksi dilakukan melalui integrasi probabilitas dari Rough Set dan similarity CBR menggunakan bobot kombinasi sebesar 0.6 dan 0.4. Pada pengujian kasus baru, diperoleh lima kasus historis paling mirip dengan similarity 0.97, empat di antaranya tergolong “Sangat Terlambat” dan satu “Terlambat”. Sementara itu, probabilitas dari Rough Set menunjukkan distribusi 0.667 untuk “Sangat Terlambat” dan 0.333 untuk “Terlambat”. Hasil integrasi memberikan skor akhir sebesar 0.720 untuk “Sangat Terlambat” dan 0.280 untuk “Terlambat”, yang menunjukkan sistem prediksi cenderung kuat terhadap kategori “Sangat Terlambat”. Pendekatan gabungan ini terbukti efektif dalam menggabungkan kekuatan generalisasi dari Rough Set dan fleksibilitas adaptif dari CBR, sehingga dapat digunakan sebagai sistem pendukung keputusan dalam evaluasi akademik berbasis data historis. Abstract The forecast of student study length is essential for academic planning, performance assessment, and recognizing possible graduation delays. This study presents a hybrid methodology that combines Rough Set theory and Case-Based Reasoning techniques to enhance the precision of predicting student study length classifications. The Rough Set approach is employed to derive classification rules from combinations of attributes, including GPA, job status, and scholarship status, as well as to compute class probabilities inside the boundary region. Simultaneously, the CBR approach is utilized to assess similarity between cases based on attribute similarity, including normalized credit hours (SKS). The prediction results are produced by integrating Rough Set probability and CBR similarity, utilizing weighted values of 0.6 and 0.4, respectively. In the test case, five historical cases with similarity scores of 0.97 were identified, four classified as “Very Late” and one as “Late”. Rough Set probability were 0.667 for “Very Late” and 0.333 for “Late”. The conclusive integrated scores were 0.720 for “Very Late” and 0.280 for “Late”, signifying that the algorithm predominantly forecasts the “Very Late” category. This hybrid methodology adeptly integrates the generalization capabilities of Rough Set theory with the adaptive versatility of Case-Based Reasoning, rendering it appropriate as a decision support system for academic assessment grounded in historical data.
English
Prediksi lama studi mahasiswa menjadi aspek penting dalam perencanaan akademik, evaluasi kinerja, serta identifikasi potensi keterlambatan kelulusan. Penelitian ini mengusulkan pendekatan hibrida dengan mengintegrasikan metode Rough Set dan Case-Based Reasoning untuk meningkatkan akurasi prediksi kelas lama studi mahasiswa. Metode Rough Set digunakan untuk mengekstraksi aturan klasifikasi berbasis kombinasi atribut IPK, status bekerja, dan status beasiswa, serta menghitung probabilitas kelas pada boundary region. Di sisi lain, metode CBR dimanfaatkan untuk menghitung similarity antar kasus berdasarkan kemiripan atribut, termasuk jumlah SKS yang dinormalisasi. Hasil prediksi dilakukan melalui integrasi probabilitas dari Rough Set dan similarity CBR menggunakan bobot kombinasi sebesar 0.6 dan 0.4. Pada pengujian kasus baru, diperoleh lima kasus historis paling mirip dengan similarity 0.97, empat di antaranya tergolong “Sangat Terlambat” dan satu “Terlambat”. Sementara itu, probabilitas dari Rough Set menunjukkan distribusi 0.667 untuk “Sangat Terlambat” dan 0.333 untuk “Terlambat”. Hasil integrasi memberikan skor akhir sebesar 0.720 untuk “Sangat Terlambat” dan 0.280 untuk “Terlambat”, yang menunjukkan sistem prediksi cenderung kuat terhadap kategori “Sangat Terlambat”. Pendekatan gabungan ini terbukti efektif dalam menggabungkan kekuatan generalisasi dari Rough Set dan fleksibilitas adaptif dari CBR, sehingga dapat digunakan sebagai sistem pendukung keputusan dalam evaluasi akademik berbasis data historis. Abstract The forecast of student study length is essential for academic planning, performance assessment, and recognizing possible graduation delays. This study presents a hybrid methodology that combines Rough Set theory and Case-Based Reasoning techniques to enhance the precision of predicting student study length classifications. The Rough Set approach is employed to derive classification rules from combinations of attributes, including GPA, job status, and scholarship status, as well as to compute class probabilities inside the boundary region. Simultaneously, the CBR approach is utilized to assess similarity between cases based on attribute similarity, including normalized credit hours (SKS). The prediction results are produced by integrating Rough Set probability and CBR similarity, utilizing weighted values of 0.6 and 0.4, respectively. In the test case, five historical cases with similarity scores of 0.97 were identified, four classified as “Very Late” and one as “Late”. Rough Set probability were 0.667 for “Very Late” and 0.333 for “Late”. The conclusive integrated scores were 0.720 for “Very Late” and 0.280 for “Late”, signifying that the algorithm predominantly forecasts the “Very Late” category. This hybrid methodology adeptly integrates the generalization capabilities of Rough Set theory with the adaptive versatility of Case-Based Reasoning, rendering it appropriate as a decision support system for academic assessment grounded in historical data.
Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.