Sistem Rekomendasi Psikotes Untuk Penjurusan Siswa Sma Menggunakan Metode Modified K-Nearest Neighbor

Mughniy, Muchlas (2017) Sistem Rekomendasi Psikotes Untuk Penjurusan Siswa Sma Menggunakan Metode Modified K-Nearest Neighbor. Sarjana thesis, Universitas Brawijaya.

Abstract

Penjurusan belajar siswa SMA merupakan pengelompokan minat belajar untuk mempermudah siswa dalam menekuni bidang ilmu dalam pendidikan tingkat lanjut. Namun penentuan potensi akademik siswa melalui bimbingan konseling membutuhkan waktu proses yang lama dengan jumlah peserta didik yang banyak. Sistem rekomendasi penjurusan bertujuan untuk memberikan rekomendasi potensi akademik siswa yang didasarkan pada potensi kemampuan kognitif menggunakan Intelligenz Struktur Test (IST). Sistem rekomendasi menggunakan metode Modified k-Nearest Neighbor (MKNN) yang mampu mengklasifikasikan potensi siswa berdasarkan kedekatan pada data training yang memiliki parameter sembilan kecerdasan kognitif dan dua kelas penjurusan sehingga didapatkan output sistem berupa rekomendasi kelas penjurusan. Berdasarkan pengujian yang dilakukan pada sistem menggunakan metode Modified k-Nearest Neighbor diperoleh rata-rata akurasi maksimum sebesar 67,95% pada 90% dataset, rata-rata akurasi sebesar 63,58% menggunakan 4-fold Cross Validation, rata-rata Sensitivity serta Specificity sebesar 23,64% dan 92,34%, perbandingan akurasi MKNN dengan KNN adalah 63,58% banding 57,11%, serta akurasi maksimum sebesar 55,26% pada reduksi hingga menjadi enam fitur menggunakan Principal Component Analysis. Sehingga sistem rekomendasi penjurusan dengan metode Modified k-

English Abstract

Major selection for high school student intended to facilitate students to focusing on specific field for higher education. However, assigning academic potential for each student through school counselor needs plenty of time. A recommendation system can generate a recommendation for major selection based on cognitive ability by Intelligenz Struktur Test (IST). Modified k-Nearest Neighbor is applied to system which classifying academic potential based on neighborhood by training data. Training data consist of nine cognitive intelligences and two majors. So, system will provide a recommendation for major. From the testing process that has been done, has obtain highest averaged accuracy on 90% dataset is 67,95%, averaged accuracy on 4-fold Cross Validation is 63,58%, averaged Sensitivity and Specificity is 23,64% and 92,34%, accuracy comparison between MKNN and KNN is 63,58% and 57,11%, and then highest accuracy for feature reduction using Principal Component Analysis is 55,26% which is reduced to six features. According to test result indicate that Modified k-Nearest Neighbor recommendation system not optimal yet to generate a recommendation for major selection.

Item Type: Thesis (Sarjana)
Identification Number: SKR/FTIK/2017/412/051706761
Uncontrolled Keywords: rekomendasi, penjurusan SMA, Modified k-Nearest Neighbor (MKNN)
Subjects: 000 Computer science, information and general works > 003 Systems
Divisions: Fakultas Ilmu Komputer > Teknik Informatika
Depositing User: Budi Wahyono Wahyono
Date Deposited: 22 Aug 2017 04:38
Last Modified: 07 Dec 2020 07:10
URI: http://repository.ub.ac.id/id/eprint/1513
[thumbnail of Muchlas Mughniy.pdf]
Preview
Text
Muchlas Mughniy.pdf

Download (4MB) | Preview

Actions (login required)

View Item View Item