Klasifikasi Status Kinerja BUMN di Indonesia menggunakan Gabungan Metode Ensemble Learning dan Synthetic Minority Oversampling Technique (SMOTE) serta mempertimbangkan Corporate Governance Indicators

Bagaskara, Affindi Mario (2023) Klasifikasi Status Kinerja BUMN di Indonesia menggunakan Gabungan Metode Ensemble Learning dan Synthetic Minority Oversampling Technique (SMOTE) serta mempertimbangkan Corporate Governance Indicators. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 6026201017-Master_Thesis.pdf] Text
6026201017-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 April 2025.

Download (4MB) | Request a copy

Abstract

Prediksi status kesehatan finansial perusahaan, yang lebih dikenal dengan istilah financial distress prediction (FDP), sangat penting dilakukan oleh perusahaan karena dapat membantu manajemen perusahaan dalam melakukan analisis kinerja perusahaan dan manajemen risiko keuangan. Dalam penelitian ini, model FDP dikembangkan menggunakan data perusahaan Badan Usaha Milik Negara (BUMN) yang terdaftar di Bursa Efek Indonesia (BEI). Model prediksi dibangun menggunakan gabungan metode ensemble learning berbasis stacking dan synthetic minority oversampling technique (SMOTE) serta mempertimbangkan corporate governance indicators (CGI). Metode ensemble learning berbasis stacking yang digunakan dalam pembangunan model prediksi melibatkan beberapa jenis base-learner seperti support vector machine, decision tree, random forest, dan extreme gradient boosting, serta sebuah meta-learner berupa sebuah pengklasifikasi berbasis logistic regression. Berbagai skenario uji coba dilakukan untuk mendapatkan model prediksi terbaik. Hasil uji coba menunjukkan bahwa penggunaan SMOTE mampu meningkatkan hasil model FDP secara signifikan. Penggunaan rasio CGI dapat memberikan peningkatan lebih lanjut dari hasil model FDP, walaupun tidak terlalu siginifikan. Secara keseluruhan penggunaan metode ensemble learning berbasis stacking mampu memberikan kinerja model FDP yang lebih baik dibandingkan dengan kinerja yang dihasilkan oleh masing-masing pengklasifikasi yang hanya melibarkan base-leaner saja. Implikasi praktis dari model yang dikembangkan dalam penelitian ini adalah tersedianya model prediksi yang dapat membantu manajemen BUMN dalam membuat keputusan berdasarkan informasi mengenai manajemen risiko keuangan mereka.
===============================================================================================================================
Prediction of the company's financial health status, which is popularly known as financial distress prediction (FDP), is crucial for companies as it can help management analyze company performance and financial risk management. In this study, we developed an FDP model using data from state-owned enterprises (SOEs) listed on the Indonesia Stock Exchange (IDX). We used a combination of stacking-based ensemble learning method and synthetic minority oversampling techniques (SMOTE), while also considering corporate governance indicators (CGI). Our stacking-based ensemble learning method used several types of base-learners, including support vector machine, decision tree, random forest, and extreme gradient boosting, as well as a meta-learner in the form of a logistic regression-based classifier. We carried out various trial scenarios to obtain the best prediction model. Our test results showed that the use of SMOTE significantly increased the FDP model's accuracy. Although the use of CGI ratio provided only a slight increase in the FDP model's performance, it was still a notable improvement. Overall, our stacking-based ensemble learning method produced a better FDP model's performance compared to the performance of each base-learner. The practical implication of our model is that it can help SOE management make informed decisions regarding their financial risk management.

Item Type: Thesis (Masters)
Uncontrolled Keywords: financial distress prediction; corporate governance indicators; ensemble learning; synthetic minority oversampling technique; stacking. ============================================================ financial distress prediction; corporate governance indicators; ensemble learning; synthetic minority oversampling technique; stacking
Subjects: T Technology > T Technology (General) > T174 Technological forecasting
T Technology > T Technology (General) > T174.5 Technology--Risk assessment.
T Technology > T Technology (General) > T385 Visualization--Technique
T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T58.6 Management information systems
T Technology > T Technology (General) > T58.62 Decision support systems
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 59101-(S2) Master Thesis
Depositing User: Affindi Mario Bagaskara
Date Deposited: 20 Feb 2023 06:31
Last Modified: 20 Feb 2023 06:31
URI: http://repository.its.ac.id/id/eprint/97638

Actions (login required)

View Item View Item