Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/123977
Title: | Implementasi Algoritma SMOTE Dalam Penanganan Imbalance Data Citra Multispektral Klasifikasi Kesuburan Sawah. |
Authors: | Priandana, Karlisa Hardhienata, Medria Kusuma Dewi Septian, Alfariz Gilang |
Issue Date: | Aug-2023 |
Publisher: | IPB University |
Abstract: | Peningkatan penggunaan teknologi terkini sangat dibutuhkan dalam dunia
pertanian khususnya untuk memenuhi kebutuhan tanaman pangan dikarenakan
peningkatan populasi manusia yang semakin cepat. Hal tersebut menyebabkan
perlunya peningkatan pemanfaatan sumber daya dan lahan yang efisien, salah satu
penerapan teknologi terkini yaitu precision fertilization (pemupukan presisi) yang
termasuk ke dalam precision farming (pertanian presisi). Namun data yang
didapatkan pada lahan sawah tidak selalu berjumlah seimbang, sehingga
mengakibatkan permasalahan yang dinamakan imbalance data. Oleh karena itu,
dibutuhkan suatu cara untuk mengatasi masalah imbalance data. Dalam penelitian
ini, masalah tersebut diatasi dengan menggunakan algoritma synthetic minority
oversampling technique (SMOTE) dan diimplementasikan pada data citra
multispektral klasifikasi kesuburan sawah. Implementasi algoritma SMOTE
membuat data seimbang dan model diimplementasi SMOTE memiliki nilai
precision, recall, dan f1-score sebesar 90%, 89%, dan 89%. Hasil ini mengungguli
hasil kinerja model tanpa menggunakan SMOTE yang memiliki nilai sebesar 63%,
45%, dan 49% dengan adanya peningkatan sebesar 27%, 44%, dan 40%. Increasing the use of the latest technology is urgently needed in the world of agriculture, especially to meet the needs of food crops due to the rapid increase in human population. This causes the need to increase the efficient use of resources and land, one of the applications of the latest technology, namely precision fertilization which is included in precision farming. However, the data obtained on paddy fields is not always balanced, resulting in a problem called data imbalance. Therefore, we need a way to overcome the problem of imbalance data. In this study, this problem was overcome by using the synthetic minority oversampling technique (SMOTE) algorithm and implemented on multispectral image data of rice field fertility classification. The implementation of the SMOTE algorithm makes the data balanced and the model implemented by SMOTE has precision, recall and f1-score values of 90%, 89%, dan 89%. These results outperform the performance results of models without using SMOTE which have values of 63%, 45%, dan 49% with an increase of 27%, 44%, dan 40%. |
URI: | http://repository.ipb.ac.id/handle/123456789/123977 |
Appears in Collections: | UT - Computer Science |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Cover.pdf Restricted Access | Cover | 118.51 kB | Adobe PDF | View/Open |
Full Text.pdf Restricted Access | Fulltext | 956 kB | Adobe PDF | View/Open |
Lampiran.pdf Restricted Access | Lampiran | 116.79 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.