Data mining untuk memprediksi prestasi siswa berdasarkan sosial ekonomi, motivasi, kedisiplinan dan prestasi masa lalu

Heri Susanto, Sudiyatno Sudiyatno

Abstract


Penelitian ini bertujuan untuk membuat prediksi prestasi belajar siswa berdasarkan status sosial ekonomi orang tua, motivasi, kedisiplinan siswa dan prestasi masa lalu menggunakan metode data mining dengan algoritma J48. Sebagai perbandingan, data penelitian dianalisis juga dengan CHAID (Chi Squared Automatic Interaction Detection) dan regresi ganda. Pendekatan penelitian yang digunakan adalah kuantitatif. Subyek penelitian ini adalah siswa tingkat X SMK Negeri 4 Surakarta berjumlah 416 siswa. Teknik pengumpulan data yang digunakan adalah dokumentasi dan angket. Hasil penelitian menunjukkan bahwa analisis prediksi menggunakan decision tree algoritma J48 memiliki akurasi sebesar 95,7%, sedangkan analisis prediksi menggunakan CHAID memiliki tingat akurasi 82,1% dan analisis regresi ganda menghasilkan tingkat signifikansi sebesar 90,6%. Berdasarkan hasil tersebut bisa disimpulkan bahwa metode J48 lebih baik dibandingkan dengan metode CHAID dan regresi ganda.

 

DATA MINING TO PREDICT STUDENT’S ACHIEVEMENT BASED ON SOCIO-ECONOMIC, MOTIVATION, DISCIPLINE AND ACHIEVEMENT OF THE PAST

Abstract

This study aims to make student achievement prediction based on socio-economic status of parents, motivation, discipline students and past achievements using data mining methods with the J48 algorithm. For comparison, the data were analyzed also with CHAID (Chi Squared Automatic Interaction Detection) and multiple regression. The research approach is quantitative. The subjects of this study were student-first level at SMK Negeri 4 Surakarta totaled 416 students. Data collection techniques used are documentation and questionnaires. The results showed that the predictive analysis using J48 decision tree algorithm has an accuracy of 95.7%, while the predictive analysis using CHAID has the rank of an accuracy of 82.1% and a multiple regression analysis resulted in a significance level of 90.6%. Based on these results it can be concluded that the J48 method is better than the CHAID and multiple regression methods.


Full Text:

PDF | 222-231


DOI: https://doi.org/10.21831/jpv.v4i2.2547

Refbacks

  • There are currently no refbacks.




Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Our journal indexed by:

ISSN 2088-2866 (print) || ISSN 2476-9401 (online)

View Journal Visitor Stats