Prediksi Fase Pertumbuhan Padi Berdasarkan Data Citra Multitemporal Landsat-8 Dengan Metode Convolutional Neural Network (CNN) Dan Rotation Forest Multiclass (Rotfor) (Studi Kasus Sampel Survei Ksa Kabupaten Poso, Provinsi Sulawesi Tengah)

Novidianto, Raditya (2021) Prediksi Fase Pertumbuhan Padi Berdasarkan Data Citra Multitemporal Landsat-8 Dengan Metode Convolutional Neural Network (CNN) Dan Rotation Forest Multiclass (Rotfor) (Studi Kasus Sampel Survei Ksa Kabupaten Poso, Provinsi Sulawesi Tengah). Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 06211950017001-Master_Thesis.pdf] Text
06211950017001-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (3MB) | Request a copy

Abstract

Pada tahun 2018 dibentuk Survei Kerangka Sampel Area (KSA) yang dilaksanakan oleh BPS untuk menghitung luas panen padi. Dalam rangka mengatasi keterbatasan KSA, maka dilakukan usulan menggunakan satelit Landsat-8 dengan menggunakan metode Machine learning (CNN dan rotfor multiclass) untuk mengklasifikasikan area panen. Kombinasi data satelit dan data offisial merupakan inovasi yang perlu dilakukan untuk mengatasi suatu keterbatasan khususnya pada Survei KSA yang dilaksanakan BPS. CNN merupakan metode yang sering digunakan dalam mendeteksi objek pada sebuah citra, sedangkan rotfor merupakan metode berbasis pohon yang memiliki kemampuan baik dalam klasifikasi dengan tipe prediktor merupakan data kontinyu dengan memanfaatkan PCA. Penelitian ini bertujuan melihat metode yang terbaik dalam memprediksi fase pertumbuhan padi. Hasil prediksi menunjukkan kinerja yang dihasilkan model CNN menghasilkan hasil yang terbaik dibandingkan dengan metode rotation forest multiclass dengan nilai sensitivity sebesar 0,9807, specificity 0,9773, accuracy 0,9766, MCC 0,9682 dan cohen kappa 0,9680 dengan G-mean sebesar 0,9742. rotfor multiclass OVO memiliki kinerja cukup baik dengan nilai nilai sensitivity sebesar 0,9106, specificity 0,9700, accuracy 0,9063, MCC 0,8760 dan cohen kappa 0,9594 dengan G-mean sebesar 0,9238. Model rotfor multiclass OVA sendiri memiliki kinerja prediksi sensitivity sebesar 0,9084, specificity 0,9699, accuracy 0,9063, MCC 0,8737 dan cohen kappa 0,9595 dengan G-mean sebesar 0,9229.
================================================================================================
In 2018 the Area Sample Framework Survey (ASF) was formed, which was carried out by BPS to calculate the rice harvested area. In order to overcome the limitations of ASF, a proposal was made to use the Landsat-8 satellite using Machine learning (CNN and rotfor multiclass) methods to classify harvested areas. The combination of satellite data and official data is an innovation that needs to be done to overcome a limitation, especially in the ASF Survey conducted by BPS. CNN is a method that is often used in detecting objects in an image, while rotfor is a tree-based method that has good ability in classification with predictor types being continuous data using PCA. This study aims to see the best method in predicting the growth phase of rice. Prediction results show that the performance generated by the CNN model produces the best results compared to the rotation forest multiclass method with a sensitivity value of 0.9807, specificity 0.9773, accuracy 0.9766, MCC 0.9682, and cohen kappa 0.9680 with a G-mean of 0.9742. Rotfor OVO multiclass has a fairly good performance with sensitivity values of 0.9106, specificity 0.9700, accuracy 0.9063, MCC 0.8760, and cohen kappa 0.9594 with a G-mean of 0.9238. In comparison, the rotfor multiclass OVA model itself has a sensitivity prediction performance of 0.9084, specificity 0.9699, accuracy 0.9063, MCC 0.8737, and cohen kappa 0.9595 with a G-mean of 0.9229.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Convolutional Neural Network, Kerangka Sampel Area, Klasifikasi, Landsat-8, Rotation Forest. Area Sampling Framework, Classification.
Subjects: H Social Sciences > HD Industries. Land use. Labor > HD108 Classification (Theory. Method. Relation to other subjects )
Q Science > Q Science (General) > Q325.5 Machine learning.
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
S Agriculture > S Agriculture (General)
Divisions: Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User: Raditya Novidianto
Date Deposited: 09 Sep 2021 08:38
Last Modified: 09 Sep 2021 08:38
URI: http://repository.its.ac.id/id/eprint/91892

Actions (login required)

View Item View Item