PREDIKSI PREFERENSI ETIKA MAHASISWA MENGGUNAKAN ALGORITMA KLASIFIKASI NAÏVE BAYES

Karima Rizqia, - (2024) PREDIKSI PREFERENSI ETIKA MAHASISWA MENGGUNAKAN ALGORITMA KLASIFIKASI NAÏVE BAYES. S1 thesis, Universitas Pendidikan Indonesia.

	Text S_TE_2001103_Title.pdf Download (5MB)
	Text S_TE_2001103_Chapter1.pdf Download (1MB)
	Text S_TE_2001103_Chapter2.pdf Restricted to Staf Perpustakaan Download (4MB)
	Text S_TE_2001103_Chapter3.pdf Download (4MB)
	Text S_TE_2001103_Chapter4.pdf Restricted to Staf Perpustakaan Download (5MB)
	Text S_TE_2001103_Chapter5.pdf Download (777kB)
	Text S_TE_2001103_Appendix.pdf Restricted to Staf Perpustakaan Download (342kB)

Abstract

Perguruan tinggi berperan penting dalam membentuk perilaku etis mahasiswa melalui pendidikan etika berkelanjutan. Dalam penelitian ini, pendekatan data mining digunakan untuk mengidentifikasi faktor-faktor yang memengaruhi perilaku etis, seperti motivasi, efikasi diri, resilience, team strain, knowledge articulation, cooperative classroom environment. Data set survei model etika mahasiswa di Yogyakarta dari Juli 2018 hingga Desember 2018 diolah dan diseimbangkan sebelum membangun prediksi dengan algoritma klasifikasi Naïve Bayes. Analisis korelasi menunjukkan bahwa knowledge articulation, motivasi, resilience, dan cooperative classroom environment signifikan memengaruhi preferensi etika mahasiswa. Dengan rasio distribusi dataset pelatihan dan pengujian 8:2, hasil klasifikasi tanpa dan dengan SMOTE mencapai akurasi 97.30% dan 85.78%. Tanpa SMOTE, 6 dari 111 sampel (5.41%) diprediksi memiliki etika rendah, sedangkan 105 dari 111 sampel (94.59%) diprediksi memiliki etika tinggi. Dengan SMOTE, 94 dari 218 sampel (43.12%) diprediksi memiliki etika rendah, dan 124 dari 218 sampel (56.88%) diprediksi memiliki etika tinggi. Namun, nilai akurasi tanpa SMOTE cenderung bias karena tingginya akurasi berasal dari prediksi yang benar pada mayoritas data kelas positif. Setelah penerapan SMOTE, akurasi turun, tetapi hasilnya lebih seimbang dan dapat dianggap baik. Universities play an important role in shaping students' ethical behavior through continuous ethics education. In this study, a data mining approach is used to identify factors that influence ethical behavior, such as motivation, self-efficacy, and classroom environment. The survey data set of ethical models of university students in Yogyakarta from July 2018 to December 2018 was processed and balanced before building the prediction with Naïve Bayes classification algorithm. Correlation analysis showed that knowledge articulation, motivation, resilience, and cooperative classroom environment significantly influenced students' ethical preferences. With a training and testing dataset distribution ratio of 8:2, the classification results without and with SMOTE achieved 97.30% and 85.78% accuracy. Without SMOTE, 6 out of 111 samples (5.41%) were predicted to have low ethics, while 105 out of 111 samples (94.59%) were predicted to have high ethics. With SMOTE, 94 out of 218 samples (43.12%) were predicted to have low ethics, and 124 out of 218 samples (56.88%) were predicted to have high ethics. However, the accuracy value without SMOTE tends to be biased because the high accuracy comes from the correct prediction on the majority of positive class data. After the application of SMOTE, the accuracy drops, but the results are more balanced and can be considered good.

Item Type:	Thesis (S1)
Additional Information:	https://scholar.google.com/citations?view_op=list_works&hl=en&authuser=3&user=YumZW58AAAAJ ID SINTA Dosen Pembimbing: Ade Gafar Abdullah: 257412
Uncontrolled Keywords:	Educational Data Mining, Naïve Bayes, SMOTE, Perilaku Etis. Ethical Behavior.
Subjects:	L Education > L Education (General) L Education > LB Theory and practice of education > LB2300 Higher Education Q Science > QA Mathematics
Divisions:	Fakultas Pendidikan Teknologi dan Kejuruan > Jurusan Pendidikan Teknik Elektro
Depositing User:	Karima Rizqia
Date Deposited:	11 Jun 2024 08:36
Last Modified:	11 Jun 2024 08:36
URI:	http://repository.upi.edu/id/eprint/118178

Actions (login required)

View Item