Please use this identifier to cite or link to this item:
https://repository.uksw.edu//handle/123456789/22061
Title: | Perbandingan Algoritma k-nn, SVM dan Random Forest untuk Klasifikasi Wilayah Risiko Kebakaran Lahan pada Data Citra Landsat 8 OLI |
Authors: | Suryoto, Evan Geraldy |
Keywords: | kebakaran lahan;K-NN;SVM;Random Forest;machine learning;IDW;penginderaan jarak jauh;Landsat 8 OLI |
Issue Date: | Nov-2020 |
Abstract: | Kebakaran lahan memiliki tingkat kerugian besar bagi yang terkena dampaknya. Dampak langsung yang dirasakan adalah adanya kabut asap yang mencemari udara di sekitar daerah kebakaran. Penelitian ini menggunakan beberapa indeks diataranya : Indeks yang dipakai antara lain ialah Enhanced Vegetation Index (EVI), Normalized Burn Ratio (NBR), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) dan Soil Adjusted Vegetation Index (SAVI). Data yang akan diteliti berupa Citra Satelit Landsat 8 OLI. Hasil analisis menggunakan korelasi pearson, rata-rata indeks memiliki korelasi kuat. Korelasi yang paling kuat ialah antara NDVI dengan SAVI dengan nilai korelasi 0,98. Hasil confusion matrix menunjukkan bahwa metode Random Forest adalah metode yang terbaik untuk penelitian ini, hal tersebut dapat dilihat dari akurasi yang bernilai 0,9995 dan Kappa yang bernilai 0,9897. Prediksi spasial menggunakan Inverse Distance Weighted (IDW) pada perhitungan yang telah dilakukan. Pengujian hubungan spasial antar kecamatan yang berpotensi kebakaran dilakukan dengan menggunakan analisis Moran's I. Land fires have a high rate of loss for those affected. The immediate impact is the presence of smog that pollutes the air around the fire area. This study uses several indexes, including The index used includes the Enhanced Vegetation Index (EVI), Normalized Burn Ratio (NBR), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) and Soil Adjusted Vegetation Index (SAVI). The data to be studied is in the form of Landsat 8 OLI Satellite Imagery. The results of the analysis using the Pearson display, the average index has a strong display. The strongest correlation is between NDVI and SAVI with a value of 0.98. The results of the confusion matrix show that the Random Forest method is the best method for this study, it can be seen from the value of 0.9995 and Kappa which is 0.9897. The spatial prediction uses Inverse Distance Weighted (IDW) in the calculations that have been done. Testing of spatial relationships between districts where fires were carried out was carried out using Moran's I analysis. |
Description: | Tidak diizinkan karya tersebut diunggah ke dalam aplikasi Repositori Perpustakaan Universitas karena telah publikasi di Journal of Computing and Modelling. |
URI: | https://repository.uksw.edu/handle/123456789/22061 |
Appears in Collections: | T1 - Informatics Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
T1_672016048_Abstract.pdf | Abstract | 150.93 kB | Adobe PDF | View/Open |
T1_672016048_Full text.pdf Restricted Access | Full text | 2.46 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.