APRILIA, LURRY PUTRI (2024) PERBANDINGAN ALGORITMA GRAPH ATTENTION NETWORK DAN CONVOLUTIONAL NEURAL NETWORK UNTUK KLASIFIKASI POTENSI HASIL PANEN TANAMAN KOPI DENGAN EKSTRAKSI FITUR VEGETATION CONDITION INDEX. S1 thesis, Universitas Mercu Buana Jakarta.
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (HAL COVER)
01 Cover.pdf Download (434kB) |
|
Text (ABSTRAK)
02 Abstrak.pdf Download (110kB) |
|
Text (BAB I)
03 Bab 1.pdf Restricted to Registered users only Download (194kB) |
|
Text (BAB II)
04 Bab 2.pdf Restricted to Registered users only Download (642kB) |
|
Text (BAB III)
05 Bab 3.pdf Restricted to Registered users only Download (150kB) |
|
Text (BAB IV)
06 Bab 4.pdf Restricted to Registered users only Download (569kB) |
|
Text (BAB V)
07 Bab 5.pdf Restricted to Registered users only Download (108kB) |
|
Text (DAFTAR PUSTAKA)
08 Daftar Pustaka.pdf Restricted to Registered users only Download (208kB) |
|
Text (LAMPIRAN)
09 Lampiran.pdf Restricted to Registered users only Download (611kB) |
Abstract
This research aims to evaluate and compare two commonly used algorithms in the context of predicting the potential harvest yield of coffee crops, namely Convolutional Neural Network (CNN) and Graph Attention Network (GAT). The primary focus of this study is to enhance prediction accuracy by integrating spatiotemporal approaches and conducting feature extraction using the Vegetation Condition Index (VCI). In experiments using CNN, the obtained results include a Cohen's Kappa Score of 0.72, an Accuracy of 0.90, Precision of 0.87, Recall of 0.90, and an F1-score of 0.88. Meanwhile, the Graph Attention Network demonstrates superior performance with a Cohen's Kappa Score of 0.82, an Accuracy of 0.93, Precision of 0.89, Recall of 0.93, and an F1-score of 0.91. These results indicate that GAT significantly outperforms CNN in modeling and understanding complex relationships in spatiotemporal data for predicting potential harvest yields. The superiority of GAT may be attributed to its ability to capture more contextual and adaptive information through attention mechanisms within the graph. Additionally, the integration of the Vegetation Condition Index as a feature extraction plays a crucial role in improving prediction accuracy. This research contributes to the development of harvest yield prediction methods by considering the spatiotemporal characteristics and specific features of coffee crops. The implications of this research can support decision-making in agriculture, enabling farmers to plan farming activities more effectively based on more accurate predictions. Keywords: Convolutional Neural Network, Graph Attention Network, Spatiotemporal, Vegetation Condition Index, Harvest Yield Prediction Penelitian ini bertujuan untuk mengevaluasi dan membandingkan dua algoritma yang umum digunakan dalam konteks prediksi potensi hasil panen tanaman kopi, yaitu Convolutional Neural Network (CNN) dan Graph Attention Network (GAT). Fokus utama penelitian ini adalah meningkatkan akurasi prediksi dengan mengintegrasikan pendekatan spatiotemporal dan melakukan ekstraksi fitur menggunakan Vegetation Condition Index (VCI). Dalam eksperimen menggunakan CNN, hasil yang diperoleh mencakup Cohen's Kappa Score sebesar 0.72, Accuracy sebesar 0.90, Precision sebesar 0.87, Recall sebesar 0.90, dan F1-score sebesar 0.88. Sementara itu, Graph Attention Network menunjukkan performa yang lebih unggul dengan Cohen's Kappa Score sebesar 0.82, Accuracy sebesar 0.93, Precision sebesar 0.89, Recall sebesar 0.93, dan F1-score sebesar 0.91. Hasil tersebut menunjukkan bahwa GAT secara signifikan mengungguli CNN dalam memodelkan dan memahami hubungan kompleks dalam data spatiotemporal untuk prediksi potensi hasil panen. Kelebihan GAT dapat disebabkan oleh kemampuannya dalam menangkap informasi yang lebih kontekstual dan adaptif melalui mekanisme attention dalam graf. Selain itu, integrasi Vegetation Condition Index sebagai fitur ekstraksi juga berperan penting dalam meningkatkan ketepatan prediksi. Penelitian ini memberikan kontribusi dalam pengembangan metode prediksi potensi hasil panen dengan mempertimbangkan karakteristik spatiotemporal dan fitur khusus tanaman kopi. Implikasi dari penelitian ini dapat mendukung pengambilan keputusan di bidang pertanian, memungkinkan para petani untuk merencanakan kegiatan pertanian dengan lebih efektif berdasarkan prediksi yang lebih akurat. Kata Kunci : Convolutional Neural Network, Graph Attention Network, Spatiotemporal, Vegetation Condition Index, Prediksi Hasil Panen.
Actions (login required)
 |
View Item |
