Analyzing the Effectiveness of Support Vector Machine and Random Forest Classifiers in Delineating the Green Area
Due to human limitations in exploring the world, the existence of remote sensing technology has made it possible and affordable for humans to study the green cover in the modern world, especially over a large region. This is so that details of the objects can be captured and monitored by satellites...
Saved in:
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Published: |
Institute of Physics
2024
|
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Due to human limitations in exploring the world, the existence of remote sensing technology has made it possible and affordable for humans to study the green cover in the modern world, especially over a large region. This is so that details of the objects can be captured and monitored by satellites or other aircraft by measuring the wavelengths of radiation that are both emitted and reflected from the area. For the past decades, various approaches have been utilized by researchers to detect green areas such as deep learning, machine learning, object based and pixel-based classification. Thus, this study aims to evaluate the effectiveness of machine learning classifiers such as Support Vector Machine (SVM) and Random Forest (RF) in detecting and delineating the green area in Universiti Teknologi Mara (UiTM) Shah Alam. Based on the study, the overall accuracy obtained by the SVM classifier is 80% with a 0.75 kappa coefficient, whereas the RF classifier managed to get 79% with a 0.74 kappa. Even though the result of both classifiers is almost the same, the accuracy of detecting green area by the SVM classifier is 93% which outperforms the RF classifier with an accuracy of 88%. This shows that the SVM classifier is more effective than the RF classifier. The detection and delineation of the green area using both machine learning approaches also showed using a map so that it is easier to be analyzed and observed. � 2023 Published under licence by IOP Publishing Ltd. |
|---|