PERFORMANCE COMPARISON BASED ON DATA MINING TECHNIQUE FOR CLASSIFICATION OF KARTU INDONESIA SEHAT (KIS) RECIPIENT

JESSICA, JESSICA (2019) PERFORMANCE COMPARISON BASED ON DATA MINING TECHNIQUE FOR CLASSIFICATION OF KARTU INDONESIA SEHAT (KIS) RECIPIENT. S1 thesis, Universitas Mercu Buana Jakarta.

Text (JURNAL MAHASISWA) Jurnal 41516010006 Jessica.pdf Restricted to Registered users only Download (1MB)

Abstract

Defects in determining of Kartu Indonesia Sehat (KIS) Recipient in Indonesia make the target audience be misplaced. The application of data mining techniques, in this case, be a solution. A comparison of classification algorithms that will be done is to determine a suitable algorithm to determine the recipient of KIS. The data used is the Integrated Data Base (BDT) Ministry of Social Affairs of the Republic of Indonesia. This research can know a suitable algorithm in the KIS recipient classification process based on the level of accuracy, precision, recall, and ROC diagram. The method used consists of 3 (three) algorithms that is C4.5, Naïve Bayes, and K-Nearest Neighbor (K-NN). This research resulted in that the K-Nearest Neighbor algorithm has an accuracy of 87.54% and is above the threshold line diagram alongside ROC C4.5 with 87.18% accuracy. Meanwhile, Naïve Bayes achieves an accuracy of 81.87% and is below the threshold. Based on the results achieved, we know that the K-Nearest Neighbor algorithm is more suitable in classifying Kartu Indonesia Sehat recipient. Key words: Data mining; C4.5; Naïve Bayes; K-NN; KIS; Kekeliruan dalam penentuan penerima Kartu Indonesia Sehat (KIS) di Indonesia membuat target peserta menjadi salah sasaran. Penerapan teknik data mining dalam kasus ini menjadi sebuah solusi. Komparasi algoritma klasifikasi yang akan dilakukan bertujuan untuk menentukan algoritma yang cocok dalam menentukan penerima KIS. Data yang digunakan adalah Basis Data Terpadu (BDT) Kementrian Sosial Republik Indonesia. Dengan penelitian ini dapat diketahui algoritma yang cocok dalam proses klasifikasi penerima KIS berdasarkan tingkat akurasi, precision, recall dan diagram ROC. Metode yang digunakan terdiri dari 3 (tiga) algoritma yaitu C4.5, Naive Bayes dan K-Nearest Neighbor(K-NN). Penelitian ini menghasilkan bahwa algoritma K-Nearest Neighbor memiliki akurasi 87,54% dan berada di atas garis threshold diagram ROC bersama C4.5 dengan accuracy 87,18%. Sedangkan Naive Bayes meraih accuracy sebesar 81,87% dan berada di bawah garis threshold. Berdasarkan hasil yang dicapai kita mengetahui bahwa algoritma K-Nearest Neighbor lebih cocok dalam pengklasifikasian penerima Kartu Indonesia Sehat. Kata kunci: Data mining; Klasifikasi; C4.5; Naive Bayes; K-NN; KIS;

Actions (login required)

View Item