Analisis Hasil Klasifikasi Tutupan Lahan Menggunakan Metode Artificial Neural Network (ANN), Support Vector Machine (SVM), Dan Random Forest (RF) Dengan Bahasa Pemrograman R.

Putri, Klarissa Ardilia (2022) Analisis Hasil Klasifikasi Tutupan Lahan Menggunakan Metode Artificial Neural Network (ANN), Support Vector Machine (SVM), Dan Random Forest (RF) Dengan Bahasa Pemrograman R. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 03311840000014_UNDERGRADUATE_THESIS.pdf Restricted to Repository staff only Download (10MB)

Abstract

Penelitian dengan data citra satelit penginderaan jauh banyak dikembangkan dengan metode klasifikasi citra. Hal ini dikarenakan hasil klasifikasi citra dapat digunakan sebagai dasaran untuk melakukan interpretasi, analisis, serta pemodelan data spasial yang dapat digunakan untuk membantu dalam penentuan berbagai kebijakan lingkungan dan sosial ekonomi. Salah satunya dikembangkan metode klasifikasi berbasis non parametrik (Non-parametric Classifier). Metode klasifikasi berbasis non parametrik yang sering digunakan adalah metode Machine Learning Decision Trees, Random Forest, Artificial Neural Networks dan Support Vector Machines. Melalui penelitian ini, dilakukan analisis perbandingan klasifikasi tutupan lahan menggunakan pendekatan Machine Learning (Artificial Neural Network, Support Vector Machine, dan Random Forest) untuk wilayah Kota Surabaya dengan menggunakan data citra satelit Landsat 8. Adapun tujuan penelitian untuk mengevaluasi performa hasil klasifikasi yang dihasilkan dari ketiga metode tersebut, sehingga perlu dilakukan analisis mengenai perbedaa hasil klasifikasi ketiga metode. Klasifikasi tutupan lahan yang diterapkan pada penelitian ini terdiri dari empat kelas yaitu, badan air, lahan terbuka, lahan terbangun, dan vegetasi. Adapun komposisi training point yang digunakan adalah 80:20, dimana 80% titik sebagai titik sample dan 20% sebagai titik validasi. Dan jumlah seluruh training point adalah 237 titik. Hasil klasifikasi tutupan lahan pada penelitian ini kemudian dilakukan uji akurasi secara kualitatif dan kuantitatif. Berdasarkan uji akurasi secara kuantitatif metode Random Forest menunjukkan hasil yang paling baik dengan nilai overal accuracy 93,33% dan kappa accuracy sebesar 91,07%. Sedangkan berdasarkan hasil uji akurasi secara kualitatif dengan menggunakan peta Rencana Detail Tata Ruang dan validasi lapangan, ketiga metode dapat mengklasifikasikan keempat kelas tutupan lahan dengan baik. Kedepannya diharapkan algoritma metode klasifikasi dapat menghasilkan jumlah kelas tutupan yang lebih banyak dan detail untuk berbagai keperluan penunjang yang lain.
===================================================================================================================================
Research with remote sensing satellite image data has been developed using image classification methods. This is because the results of image classification can be used as a basis for interpreting, analyzing, and modeling spatial data that can be used to assist in determining various environmental and socio-economic policies. One of them developed a classification method based on a non-parametric classifier. Non-parametric-based classification methods that are often used are Machine Learning Decision Trees, Random Forests, Artificial Neural Networks, and Support Vector Machines. Through this research, a comparative analysis of land cover classification was carried out using a Machine Learning approach (Artificial Neural Network, Support Vector Machine, and Random Forest) for Surabaya City using Landsat 8 satellite imagery data. This research aims to evaluate the performance of the classification results from the three methods, so it is necessary to analyze the differences in the results of the three classification methods. The land cover classification applied in this study consisted of four classes, there are water, bare land, urban land, and vegetation. The composition of the training points used is 80:20, which 80% points as sample points, and 20% as validation points. And the total number of training points is 237 points. The land cover classification results in this study were tested for accuracy quantitatively and qualitatively. Based on the quantitative accuracy test, the Random Forest method showed the best results with an overall accuracy value of 93.33% and a kappa accuracy of 91.07%. Meanwhile, based on the results of the qualitative accuracy test using the Detailed Spatial Plan map and field validation, the three methods can classify the four land cover classes well. In the future, it is hoped that the classification method algorithm can produce a more significant number of cover classes and details for various other supporting purposes.

Item Type:	Thesis (Other)
Additional Information:	RSG 333.731 3 Put a-1 2022
Uncontrolled Keywords:	Artificial Neural Network, Machine Learning, Random Forest, Support Vector Machine. Artificial Neural Network, Machine Learning, Random Forest, Support Vector Machine.
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29202-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	18 May 2026 08:15
Last Modified:	18 May 2026 08:15
URI:	http://repository.its.ac.id/id/eprint/133232

Actions (login required)

View Item