Prediksi Financial Distress pada Perusahaan Sektor Consumer Cyclicals di Indonesia Menggunakan Pendekatan Artificial Neural Network dan Extreme Gradient Boosting

Putri, Nadya Anastasya Eka (2026) Prediksi Financial Distress pada Perusahaan Sektor Consumer Cyclicals di Indonesia Menggunakan Pendekatan Artificial Neural Network dan Extreme Gradient Boosting. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043221093-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk menganalisis dan memprediksi kondisi financial distress pada perusahaan sektor consumer cyclicals di Indonesia selama periode 2020-2024 menggunakan pendekatan Artificial Neural Network (ANN) dan Extreme Gradient Boosting (XGBoost). Hasil analisis menunjukkan bahwa jumlah perusahaan yang mengalami financial distress tertinggi terjadi pada tahun 2020 dengan total 77 perusahaan kemudian cenderung menurun pada tahun-tahun berikutnya. Pengujian model prediksi menunjukkan bahwa ANN memiliki kinerja yang lebih unggul dibandingkan XGBoost dengan tingkat akurasi sebesar 89,19% pada arsitektur Multilayer Perceptron (6-12-1), learning rate 0,003, batch size 16, dan epoch 800. Sementara itu, model XGBoost menghasilkan akurasi sebesar 84,68% dengan parameter N estimator 30, learning rate 0,03, max depth 5, subsample 0,8, dan colsample by tree 0,5. ANN juga menunjukkan sensitivitas dan spesifisitas yang lebih baik dalam mengenali perusahaan yang mengalami kesulitan keuangan. Hasil analisis SHAP menunjukkan nilai SHAP rata-rata ROA sebesar 0,200, NPM mendekati 0,175, dan quick ratio sebesar 0,075 menjadi tiga variabel paling dominan dalam prediksi financial distress dengan rasio profitabilitas yang memiliki pengaruh lebih kuat dibandingkan rasio likuiditas. Sebagai bentuk implementasi, model prediksi dikembangkan ke dalam aplikasi berbasis web menggunakan framework Streamlit yang memungkinkan pengguna memperoleh hasil klasifikasi kondisi keuangan perusahaan secara langsung, sehingga dapat dimanfaatkan sebagai sistem peringatan dini (early warning system).
============================================================================================================================
This study aims to analyze and predict financial distress conditions among consumer cyclicals sector companies in Indonesia during the period 2020–2024 using Artificial Neural Network (ANN) and Extreme Gradient Boosting (XGBoost) approaches. The results indicate that the highest number of financially distressed companies occurred in 2020, with a total of 77 firms, followed by a declining trend in subsequent years. Model performance evaluation shows that ANN outperforms XGBoost, achieving an accuracy of 89.19% using a Multilayer Perceptron architecture (6-12-1), a learning rate of 0.003, a batch size of 16, and 800 epochs. Meanwhile, the XGBoost model attains an accuracy of 84.68% with 30 estimators, a learning rate of 0.03, a maximum depth of 5, a subsample rate of 0.8, and a colsample by tree of 0.5. In addition to higher accuracy, ANN demonstrates superior sensitivity and specificity in identifying companies experiencing financial difficulties. SHAP analysis shows that the average SHAP value of ROA is 0.200, NPM is close to 0.175, and the quick ratio is 0.075, making these three variables the most dominant in predicting financial distress, with profitability ratios having a stronger influence than liquidity ratios. As an implementation, the predictive model is deployed into a web-based application using the Streamlit framework, allowing users to input financial ratios and obtain real-time financial distress classifications, thereby serving as an early warning system.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Artificial Neural Network, Consumer Cyclicals, Extreme Gradient Boosting, Financial Distress Artificial Neural Network, Consumer Cyclicals, Extreme Gradient Boosting, Financial Distress
Subjects:	Q Science Q Science > Q Science (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Nadya Anastasya Eka Putri
Date Deposited:	29 Jan 2026 04:18
Last Modified:	29 Jan 2026 04:18
URI:	http://repository.its.ac.id/id/eprint/131005

Actions (login required)

View Item