Akses terbuka Jurnal SINTA 2
Penerapan Data Mining untuk Prediksi Financial Distress pada Perusahaan Sektor Keuangan Periode 2020-2023
Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK) · Vol. 13 No. 5 (2026) · hlm. 1275–1286
- ISSN
- 2528-6579, 2355-7699
- Terbit
- 2026-11-02
- Penerbit
- Fakultas Ilmu Komputer, Universitas Brawijaya
- Bahasa
- id
Abstrak
Bahasa Indonesia
Penelitian ini mengkaji performa model prediksi financial distress pada perusahaan sektor keuangan di Bursa Efek Indonesia (BEI) selama periode 2020-2023, menggunakan metode data mining berbasis algoritma random forest, serta mengidentifikasi rasio-rasio keuangan paling signifikan dalam klasifikasi. Dampak pandemi Covid-19 terhadap kinerja sektor keuangan menjadikan kemampuan prediksi dini semakin krusial bagi pemangku kepentingan. Penelitian ini menganalisis laporan keuangan tahunan 87 perusahaan (348 sampel) dengan delapan rasio keuangan sebagai variabel independen, membandingkan tiga proporsi pembagian data, dan menerapkan teknik SMOTE untuk mengatasi ketidakseimbangan kelas. Model dengan proporsi data 60:20:20 setelah penerapan SMOTE memberikan performa optimal, baik pada data validasi (akurasi 92,9%, precision 86,5%, recall 83,4%, F1-score 84,8%) maupun data uji (akurasi 92,9%, precision 87,5%, recall 84,7%, F1-score 86,0%). Analisis atribut mengidentifikasi return on assets, net profit margin, total asset turnover, dan working capital to total assets ratio sebagai indikator paling signifikan dalam membedakan perusahaan distress dan non-distress, berdasarkan bobot dalam random forest dan posisi dalam struktur pohon keputusan J48. Penelitian ini memberikan kontribusi penting terhadap pengembangan model prediksi financial distress yang relevan untuk sektor keuangan Indonesia selama periode pandemi, serta menunjukkan efektivitas pendekatan SMOTE dan variasi proporsi data dalam menangani dataset terbatas dan tidak seimbang. Temuan ini memberikan implikasi praktis bagi manajemen, regulator, dan investor dalam memperkuat sistem peringatan dini dan pengambilan keputusan strategis. Abstract This study examines the performance of financial distress prediction models for financial sector companies listed on the Indonesia Stock Exchange (IDX) during the 2020-2023 period, using data mining methods based on the random forest algorithm and identifying the most significant financial ratios in the classification process. The impact of the Covid-19 pandemic on the financial sector’s performance has made early prediction capabilities increasingly critical for stakeholders. This study analyzes the annual financial statements of 87 companies (348 samples), using eight financial ratios as independent variables, comparing three data split proportions, and applying the SMOTE technique to address class imbalance. The model employing a 60:20:20 data split after SMOTE implementation demonstrated optimal performance, both on the validation set (accuracy 92.9%, precision 86.5%, recall 83.4%, F1-score 84.8%) and the test set (accuracy 92.9%, precision 87.5%, recall 84.7%, F1-score 86.0%). Attribute analysis identified return on assets, net profit margin, total asset turnover, and working capital to total assets ratio as the most significant indicators in distinguishing between distressed and non-distressed companies, based on their weight in the random forest model and their position within the J48 decision tree structure. This study provides an important contribution to the development of relevant financial distress prediction models for Indonesia’s financial sector during the pandemic period and demonstrates the effectiveness of the SMOTE approach and data split variations in handling limited and imbalanced datasets. The findings offer practical implications for management, regulators, and investors in strengthening early warning systems and supporting strategic decision-making.
Inggris
Penelitian ini mengkaji performa model prediksi financial distress pada perusahaan sektor keuangan di Bursa Efek Indonesia (BEI) selama periode 2020-2023, menggunakan metode data mining berbasis algoritma random forest, serta mengidentifikasi rasio-rasio keuangan paling signifikan dalam klasifikasi. Dampak pandemi Covid-19 terhadap kinerja sektor keuangan menjadikan kemampuan prediksi dini semakin krusial bagi pemangku kepentingan. Penelitian ini menganalisis laporan keuangan tahunan 87 perusahaan (348 sampel) dengan delapan rasio keuangan sebagai variabel independen, membandingkan tiga proporsi pembagian data, dan menerapkan teknik SMOTE untuk mengatasi ketidakseimbangan kelas. Model dengan proporsi data 60:20:20 setelah penerapan SMOTE memberikan performa optimal, baik pada data validasi (akurasi 92,9%, precision 86,5%, recall 83,4%, F1-score 84,8%) maupun data uji (akurasi 92,9%, precision 87,5%, recall 84,7%, F1-score 86,0%). Analisis atribut mengidentifikasi return on assets, net profit margin, total asset turnover, dan working capital to total assets ratio sebagai indikator paling signifikan dalam membedakan perusahaan distress dan non-distress, berdasarkan bobot dalam random forest dan posisi dalam struktur pohon keputusan J48. Penelitian ini memberikan kontribusi penting terhadap pengembangan model prediksi financial distress yang relevan untuk sektor keuangan Indonesia selama periode pandemi, serta menunjukkan efektivitas pendekatan SMOTE dan variasi proporsi data dalam menangani dataset terbatas dan tidak seimbang. Temuan ini memberikan implikasi praktis bagi manajemen, regulator, dan investor dalam memperkuat sistem peringatan dini dan pengambilan keputusan strategis. Abstract This study examines the performance of financial distress prediction models for financial sector companies listed on the Indonesia Stock Exchange (IDX) during the 2020-2023 period, using data mining methods based on the random forest algorithm and identifying the most significant financial ratios in the classification process. The impact of the Covid-19 pandemic on the financial sector’s performance has made early prediction capabilities increasingly critical for stakeholders. This study analyzes the annual financial statements of 87 companies (348 samples), using eight financial ratios as independent variables, comparing three data split proportions, and applying the SMOTE technique to address class imbalance. The model employing a 60:20:20 data split after SMOTE implementation demonstrated optimal performance, both on the validation set (accuracy 92.9%, precision 86.5%, recall 83.4%, F1-score 84.8%) and the test set (accuracy 92.9%, precision 87.5%, recall 84.7%, F1-score 86.0%). Attribute analysis identified return on assets, net profit margin, total asset turnover, and working capital to total assets ratio as the most significant indicators in distinguishing between distressed and non-distressed companies, based on their weight in the random forest model and their position within the J48 decision tree structure. This study provides an important contribution to the development of relevant financial distress prediction models for Indonesia’s financial sector during the pandemic period and demonstrates the effectiveness of the SMOTE approach and data split variations in handling limited and imbalanced datasets. The findings offer practical implications for management, regulators, and investors in strengthening early warning systems and supporting strategic decision-making.
Kata kunci
Jaringan kolaborasi para penulis
40 penulis · 93 tautan kepenulisan bersama · 20 lintas organisasi · menampilkan 40 yang paling banyak terhubung
Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.
- University of Brawijaya
- Poltekkes Kemenkes Kendari
- Universitas Muhammadiyah Malang
- State University of Semarang
- Institut Teknologi Sepuluh Nopember
- National Research and Innovation Agency
- Indonesian Institute of Sciences
- Afiliasi tidak diketahui
- Organisasi berbeda
- Organisasi sama
- Afiliasi tidak diketahui
- Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
| Rekan penulis | Organisasi | Artikel bersama |
|---|---|---|
| Admaja Dwi Herlambang | University of Brawijaya | 5 |
| Satrio Agung Wicaksono | University of Brawijaya | 4 |
| Nanang Yudi Setiawan | University of Brawijaya | 4 |
| Yusi Tyroni Mursityo | University of Brawijaya | 3 |
| Dian Eka Ratnawati | University of Brawijaya | 3 |
| Fahrur Rozi | State University of Semarang | 2 |
| Septiyan Andika Isanta | Universitas Muhammadiyah Malang | 2 |
| Aditya Rachmadi | University of Brawijaya | 2 |
| Faizatul Amalia | University of Brawijaya | 2 |
| Fatmawati Fatmawati | University of Brawijaya | 2 |
| Tri Afirianto | University of Brawijaya | 2 |
| Yustiari Yustiari | Poltekkes Kemenkes Kendari | 2 |
| Aryo Pinandito | University of Brawijaya | 2 |
| Diana Purwitasari | Institut Teknologi Sepuluh Nopember | 1 |
| Yufis Azhar | Universitas Muhammadiyah Malang | 1 |
| Cindy Felita Nur Alimah | University of Brawijaya | 1 |
| Muhammad Rosyid Mubarok | University of Brawijaya | 1 |
| Anisah Ikhsawiyanthi | University of Brawijaya | 1 |
| Pratama Aprilianto | University of Brawijaya | 1 |
| Hanifah Muslimah Az-Zahra | University of Brawijaya | 1 |
| Alfansya Achmad Mulyadi | University of Brawijaya | 1 |
| Meilita Dwi Aryati | University of Brawijaya | 1 |
| Azka Azkiyah | University of Brawijaya | 1 |
| Ahmad Afif Supianto | University of Brawijaya | 1 |
| Abdul Harris Wicaksono | University of Brawijaya | 1 |
| Didik Krisnandi | National Research and Innovation Agency | 1 |
| Ana Heryana | Indonesian Institute of Sciences | 1 |
| M. Ro'if | University of Brawijaya | 1 |
| Rihadatul 'Aisy | University of Brawijaya | 1 |
| Dion Ricky Saputra | University of Brawijaya | 1 |
| Nandana Rifqi Irfansyah | Afiliasi tidak diketahui | 1 |
| Shafa Nathaniela Salwa | Afiliasi tidak diketahui | 1 |
| Uun Hariyanti | Afiliasi tidak diketahui | 1 |
| Indana Zulfa | Afiliasi tidak diketahui | 1 |
| Okta Purnawirawan | University of Brawijaya | 1 |
| Maurish Sofie Rahmi Batita | Afiliasi tidak diketahui | 1 |
Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.