Klasifikasi Citra Umbi Menggunakan Metode Support Vector Machine dan Decision Tree

AMPUR, Paskalis (2025) Klasifikasi Citra Umbi Menggunakan Metode Support Vector Machine dan Decision Tree. Undergraduate thesis, Universitas Katolik Widya Mandira Kupang.

	Text ABSTRAK.pdf Download (706kB)
	Text BAB I.pdf Download (231kB)
	Text BAB II.pdf Restricted to Repository staff only Download (458kB)
	Text BAB III.pdf Restricted to Repository staff only Download (407kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (546kB)
	Text BAB V.pdf Restricted to Repository staff only Download (544kB)
	Text BAB VI.pdf Download (237kB)
	Text DAFTAR PUSTAKA DAN SURAT KETERANGAN HASIL CEK PLAGIASI.pdf Download (283kB)

Abstract

Tubers are an important agricultural commodity that contributes to food diversity and nutritional adequacy of communities, but challenges arise in identifying different types of tubers that have physical similarities, such as potatoes and white sweet potatoes, which can confuse farmers. This is what is used as a basic reference in this study, the researcher will design a classification system in the field of agriculture that is very large by making an image processing model based on artificial intelligence. This study aims to identify the types of tubers using the Support Vector Machine (SVM) and Decision Tree methods, with evaluation using a confusion matrix on a dataset of 1000 samples from yams, potatoes, yellow sweet potatoes, purple sweet potatoes, and cassava. This study uses the classification methods of Support Vector Machine and Decision Tree as well as Orange data mining as a tool to identify the type of tubers. The results show that the SVM method achieves 99.8% accuracy at 2-fold cross validation and remains stable up to 20-fold, while Decision Tree initially only reaches 82.7% but significantly improves to 88% at 5-fold. This study shows that the SVM method is superior in identifying tuber types, and the increase in the number of folds has a positive impact on the Decision Tree method.

Actions (login required)

View Item