Klasifikasi Sustainability-Adjusted Rating berdasarkan Indeks SolAbility dengan Metode BPNN dan SAMME

Wiyata, Rachel Cahyani Ardhika (2024) Klasifikasi Sustainability-Adjusted Rating berdasarkan Indeks SolAbility dengan Metode BPNN dan SAMME. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006201100-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (5MB) | Request a copy

Abstract

Obligasi negara merupakan instrumen keuangan dengan skema, yaitu pemerintah diwajibkan membayar kembali pinjaman secara berkala dalam periode tertentu kepada pemberi pinjaman dengan jumlah yang sudah ditambahkan dengan bunga. Kemampuan dan kelayakan suatu pemerintah dalam memenuhi kewajiban membayar hutangnya dapat dinilai melalui suatu peringkat kredit yang disebut dengan sovereign rating. Sebagian besar pemeringkat kredit akan memberikan credit rating yang rendah kepada negara berkembang. SolAbility merupakan konsultan manajemen yang melakukan penelitian terkait ESG dan menerbitkan sustainability-adjusted rating. Sustainability-adjusted rating merupakan credit rating yang mempertimbangkan aspek lingkungan, sosial, tata kelola, dan ekonomi sehingga dapat menganalisis lebih dalam terkait risiko pada obligasi. Penelitian ini akan berfokus pada klasifikasi yang menggunakan metode backpropagation neural network (BPNN) dan Ssagewise additive modeling using a multi-class exponential loss function (SAMME). Kategori sustainability-adjusted rating terbagi atas tiga kategori utama, yaitu low, medium, dan high. Data sustainability-adjusted rating yang digunakan adalah tahun 2022 – 2023. Model BPNN dan SAMME terbaik diperoleh melalui proses GridSearchCV yang menghasilkan nilai F_1 score tertinggi. Hasil dari penelitian menunjukkan bahwa model terbaik adalah model SAMME pada data yang tidak diterapkan SMOTE dengan keakuratan macro F_1 score data testing adalah 80,6%. Model SAMME terbaik didapatkan pada data tanpa SMOTE dengan kombinasi hyperparameter yang terdiri atas max depth sebesar 1, n_estimator sebanyak 50, dan learning rate sebesar 1.

===================================================================================================================================
Government bonds are financial instruments structured such that the government is obliged to periodically repay the loan over a certain period to the lender, with the amount repaid including interest. The ability and feasibility of a government to meet its debt repayment obligations can be assessed through a credit rating known as a sovereign rating. Most credit rating agencies tend to assign lower credit ratings to developing countries. SolAbility is a management consultancy that conducts research related to Environmental, Social, and Governance (ESG) factors and publishes sustainability-adjusted ratings. A sustainability-adjusted rating is a credit rating that considers environmental, social, governance, and economic aspects, providing a deeper analysis of the risks associated with bonds. This research focuses on classification using two methods: backpropagation neural network (BPNN) and Stagewise Additive Modeling using a Multi-class Exponential loss function (SAMME). The sustainability-adjusted rating categories are divided into three main categories: High, medium, and low. The sustainability-adjusted rating data used spans the years 2022 to 2023. The best models for BPNN and SAMME were obtained through the GridSearchCV process, which yielded the highest F_1 score. The findings demonstrate that the optimal model is the SAMME model applied to data without SMOTE, achieving a macro F_1 score of 80.6% on the testing data. The best SAMME model was identified on data without SMOTE using a hyperparameter configuration of a max depth of 1, 50 estimators, and a learning rate of 1.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Pembangunan Keberlanjutan, ESG, Sustainability-Adjusted Rating, BPNN, SAMME, Sustainable Development, ESG, Sustainability-Adjusted Rating, BPNN, SAMME
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA9.58 Algorithms
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Rachel Cahyani Ardhika Wiyata
Date Deposited:	08 Jul 2024 07:34
Last Modified:	08 Jul 2024 07:36
URI:	http://repository.its.ac.id/id/eprint/108185

Actions (login required)

View Item