Hilmiyah, Fathin (2017) Prediksi Kinerja Mahasiswa menggunakan Support Vector Machine untuk Pengelola Program Studi di Perguruan Tinggi (Studi Kasus: Program Studi Magister Statistika ITS). Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
9115205311-Master_Thesis.pdf - Published Version Download (2MB) | Preview |
Abstract
Kesuksesan sebuah perguruan tinggi sering dilihat dari keberhasilan prestasi belajar mahasiswa yang bernaung dalam institusi tersebut. Salah satu langkah yang dibutuhkan yakni dengan melakukan prediksi kinerja akademik bagi calon mahasiswa. Prediksi kinerja mahasiswa ini dapat digunakan sebagai salah satu penunjang keputusan evaluasi kinerja mahasiswa. Program Studi Magister Statistika ITS Surabaya belum memiliki tools untuk memprediksi kinerja dan keberhasilan tersebut. Maka dari itu, peneliti melakukan penelitian untuk membuat pemodelan Prediksi Kinerja Mahasiswa untuk Pengelola Program Studi di Perguruan Tinggi (Studi Kasus: Program Studi Magister Statistika ITS). Penelitian ini menggunakan data akademik berupa sampel data mahasiswa Magister Statistika ITS sejumlah 318 data. Metode yang digunakan adalah metode Support Vector Machine (SVM). Pelatihan dan pengujian sistem dilakukan dengan metode 10-fold Cross Validation dengan mengukur hasil akurasi, presisi dan recall. Berdasarkan uji performa perbandingan fungsi Kernel, Linear Kernel merupakan fungsi yang paling cocok untuk menghasilkan prediksi paling optimal. Dari uji coba yang dilakukan, penerapan SVM memiliki hasil perhitungan akurasi, presisi dan recall yang lebih baik bila dibandingkan dengan penggunaan metode Regresi Logistik. Berdasarkan uji pengaruh setiap variabel, performansi kinerja mahasiswa dapat ditingkatkan secara signigikan dengan mempertimbangkan beberapa indikator mahasiswa yakni asal daerah, status kerja, jalur masuk, nilai toefl, nilai TPA, akreditasi asal institusi, IPK, lama studi dan waktu tunggu mahasiswa. =================================================================
The success of any higher educational institutions is often seen from the success of their student achievement. One of the steps needed is to predict academic performance for students. These student performance predictions can be used to improve the quality of managerial decisions and to impart quality education. Magister Study Program at Department of Statistics ITS Surabaya does not have the tools to predict the performance of their students yet. Therefore, researchers conduct this paper to make modeling of Prediction Student Performance for Management of Higher Education at Magister Study Program, Department of Statistics ITS using Support Vector Machine (SVM). The dataset comprises of 318 student records that provided information about student demographics and previous academic standings in which 12 significant variable of students were extracted for experimentation in the study. Training and system testing is done by 10-fold Cross Validation method by measuring accuracy, precision and recall. Based on comparative performance Kernel function for SVM, the result show that Linear Kernel is the most suitable Kernel function to produce the most optimal prediction for student performance. From the experiments performed, the application of SVM has a better accuracy, precision and recall calculation results when compared with the use of logistic regression method. Based on the test of the influence of each variable, the quality of student performance can be improved significantly by considering some student indicator they are work status, toefl score, TPA score, accreditation of institution before, GPA, study duration and waiting time of student from graduate to join magister program.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Prediksi, Mahasiswa, Support Vector Machine, Prediction, Student Performance |
| Subjects: | T Technology > T Technology (General) > T58.62 Decision support systems |
| Divisions: | Faculty of Business and Management Technology > Management Technology |
| Depositing User: | Fathin Hilmiyah |
| Date Deposited: | 07 Dec 2017 08:09 |
| Last Modified: | 06 Mar 2019 03:59 |
| URI: | http://repository.its.ac.id/id/eprint/46712 |
Actions (login required)
 |
View Item |