Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/71912
Title: Deteksi pencilan data titik api di provinsi riau menggunakan algoritme Clustering K-Means
Other Titles: Outlier detection on hotspot data in province riau using K-Means Clustering algorithm
Authors: Sitanggang, Imas Sukaesih
Baehaki, Dhiya Aulia Muhamad
Issue Date: 2014
Publisher: Bogor Agricultural University (IPB)
Abstract: Kebakaran hutan merupakan masalah yang selalu berulang di Provinsi Riau. Salah satu solusi dari masalah ini adalah memanfaatkan data titik api hasil penginderaan jarak jauh. Tujuan penelitian ini adalah mendeteksi pencilan yang terdapat pada data titik api di Provinsi Riau dari tahun 2001 hingga 2012. Deteksi pencilan dilakukan dengan menggunakan metode clustering k-means serta digunakan pendekatan pencilan global dan kolektif. Pada fungsi kmeans digunakan nilai k sebesar 10 sehingga dihasilkan nilai sum of squared error sebesar 18 526.14. Hasil menunjukkan bahwa berdasarkan pendekatan pencilan kolektif dideteksi pencilan titik api terdapat pada cluster 5, 7, dan 10. Di samping itu diperoleh 30 pencilan titik api berdasarkan pendekatan pencilan global. Kemunculan pencilan pada data titik api banyak terjadi pada bulan Februari, Maret, Juni, dan Agustus. Frekuensi pencilan titik api tertinggi terjadi pada tahun 2005 yaitu sebanyak 1118 titik api pada 21 Juni 2005. Adapun frekuensi pencilan titik api terkecil yaitu sebanyak 295 titik api pada tahun 2005 dan rata-rata frekuensi pencilan titik api sebesar 482.22.
Forest fire is considered as periodic event in several areas in Indonesia including in Riau Province. One of solutions in fire prevention is analyzing hostpot occurrences data that are produced by remote sensing technology The objective of this research is to detect outliers on hotspot data in Riau Province for the period 2001 to 2012. Outlier detection was done using the clustering k-means algorithm, and use a approach global outliers and collective outlier. The best clustering result was selected on the number of cluster 10 and the sum of squared error value is 18526.14. The results showed that based on a collective outlier approach, the outliers were obtained on cluster 5, 7 and 10. In addition, 30 outliers on hotspot data were detected based on the global outlier approach. The outliers on the hotspot data were mostly occurred in February, March, June, and August. The highest frequency of outliers occurred in 2005 reach 1118 hotspots on 21 June 2005. While the lowest frequency of outliers is 295 in 2005 and the average frequency of outliers is 482.22. Keywords: clustering, hotspot, k-means, outlier, riau
URI: http://repository.ipb.ac.id/handle/123456789/71912
Appears in Collections:UT - Computer Science

Files in This Item:
File Description SizeFormat 
G14dam.pdf
  Restricted Access
Full Text1.14 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.