Rancang Bangun Aplikasi Data Mining untuk Memprediksi Hasil Belajar Siswa Sekolah Menengah Atas Berbasis Web dengan Algoritma K-NN (Studi Kasus: SMKN 2 Pekanbaru)
Abstrak: Curriculum 2013
(K-13) was first announced in 2014 which has been applied to number of schools.
Preparation of this new curriculum by the government aimed at making education
in Indonesia is not only focused on cognitive aspects or skills possessed, but
also at students' interest and motivation. Unfortunately, behind the goal,
there are issues occured in the school during the application of K-13. Those
are input process and values conversion that takes relatively much time. The
things are caused by the dissimilarity of the standards and the assessment
scale between current curriculum with the previous one. Meanwhile, the academic
system running in schools is still pretty conventional. Therefore, this
research will construct an application which have capability to handle the
things. Beside those additional features, this research is build an application
in order to apply the data mining with k-NN algorithm to predict students
learning outcomes based on certain subjects. Data source that used in this
research were consisted into 500 data training that covered up all classes or
labels. Testing methods which have been applied are black box testing and
confusion matrix. There are 3 techniques of black box testing that applied in
order to test the system functionality according to its input values. Those are
equivalence class partitioning, boundary value analysis and decision table
based testing. Meanwhile in confusion matrix, it has been done 3 times testing
according by k value in k-NN algorithm. With k-5 acquired accurate rate 79.34%,
k-10 with accurate rate 62.67%, then k-15 with accurate rate 64%. Thus,
information that can conluded from those testing methods is the algorithm with
k-5 is more accurate than any others.
Kata Kunci: Web Application,
Data Mining, k-NN, Classification, Prediction, Academic Achievement
Penulis: Okta Riveranda,
Muhammad Ihsan Zul, Maksum Rois Adin Saf
Kode Jurnal: jptkomputerdd160557
