Please use this identifier to cite or link to this item:
https://repository.uksw.edu//handle/123456789/35205
Title: | Implementasi Machine learning dalam Prediksi Kekeringan Berdasarkan Data Citra Landsat 8 OLI di Purbalingga (2020-2023) |
Authors: | Simamora, Anita Nilam Agatha Nauli |
Keywords: | Kekeringan;Indeks Vegetasi;Random Forest;XGBoost;Machine learning |
Issue Date: | 28-Jun-2024 |
Abstract: | Penelitian ini mengeksplorasi penerapan machine learning untuk memprediksi kekeringan di Kabupaten Purbalingga menggunakan data citra Landsat 8 OLI dari tahun 2020 hingga 2023. Metode XGBoost dan Random Forest dipilih karena kemampuan mereka dalam menangani data kompleks dan memberikan prediksi akurat. Data penginderaan jauh dari Landsat 8 OLI dikombinasikan dengan berbagai indeks vegetasi seperti NDVI, SAVI, dan EVI. Hasil analisis menunjukkan bahwa metode XGBoost mencapai akurasi 86,13% dengan nilai Kappa 82,66%, sedangkan Random Forest mencapai akurasi 84,62% dengan nilai Kappa 80,78%. NDVI terbukti menjadi variabel dominan dalam prediksi kekeringan. Implementasi model ini memberikan estimasi akurat dan dapat diandalkan untuk mengidentifikasi kondisi kekeringan, penting untuk perencanaan dan mitigasi risiko. Studi ini menunjukkan bahwa integrasi data penginderaan jauh dengan algoritma machine learning dapat meningkatkan efisiensi pengelolaan irigasi dan penggunaan lahan.
Kata kunci: Kekeringan, Machine learning, XGBoost, Random Forest, Indeks Vegetasi This research explores the application of machine learning to predict drought in Purbalingga Regency using Landsat 8 OLI image data from 2020 to 2023. The XGBoost and Random Forest methods were chosen for their ability to handle complex data and provide accurate predictions. Remote sensing data from Landsat 8 OLI was combined with various vegetation indices such as NDVI, SAVI, and EVI. The analysis results showed that the XGBoost method achieved 86.13% accuracy with a Kappa value of 82.66%, while Random Forest achieved 84.62% accuracy with a Kappa value of 80.78%. NDVI proved to be the dominant variable in drought prediction. The implementation of this model provides accurate and reliable estimates to identify drought conditions, important for planning and risk mitigation. This study shows that the integration of remote sensing data with machine learning algorithms can improve the efficiency of irrigation and land use management. Keywords: Drought, Machine learning, XGBoost, Random Forest, Vegetation Index |
URI: | https://repository.uksw.edu//handle/123456789/35205 |
Appears in Collections: | T1 - Informatics Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
T1_672017279_Judul.pdf | 694.53 kB | Adobe PDF | View/Open | |
T1_672017279_Isi.pdf | 698.18 kB | Adobe PDF | View/Open | |
T1_672017279_Daftar Pustaka.pdf | 319.71 kB | Adobe PDF | View/Open | |
T1_672017279_FormulirPenyerahanLisensiNoneksklusifDanPilihanEmbargoTugasTalentaUnggul.pdf Until 9999-01-01 | 988.89 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.