ANALISIS PREDIKSI POTENSI KEBANGKRUTAN DENGAN METODE SUPPORT VECTOR MACHINE (SVM) PADA PERUSAHAAN YANG TERDAFTAR DI BURSA EFEK INDONESIA (BEI) YANG TERKENA SUSPENSI TAHUN 2024

This study aims to analyze the potential bankruptcy of companies listed on the Indonesia Stock Exchange (IDX) and suspended in 2024 using the Support Vector Machine (SVM) method, while identifying the most influential financial ratios and measuring the model's accuracy in classifying bankrupt and non-bankrupt companies. This quantitative study used a sample of 20 companies with 54 observational data from the 2021–2023 period through matched-pair sampling, then analyzed using Polykernel SVM and 10-fold crossvalidation. The variables used include Working Capital/Total Assets, Retained Earnings/Total Assets, EBIT/Total Assets, and Book Value of Equity/Book Value of Debt as the main indicators of a company's financial condition, while the dependent variable is bankruptcy or financial distress status. The results showed that the SVM model had an accuracy of 81.48% with a Kappa value of 0.6296, an average precision of 0.845, an average recall of 0.815, an average F-measure of 0.811 and Area under Curve (AUC/ROC) 0.815 . The largest coefficient is found in the Book Value of Equity/Book Value of Debt ratio of -3.0815, indicating that Leverage ratio is the most dominant factor in bankruptcy
prediction. Furthermore, the model is able to identify distress classes with a high recall of 0.963. Theoretically, this study strengthens the literature on the effectiveness of machine
learning, particularly SVM, in predicting potential bankruptcy through financial ratios. Practically, the results of this study can be used as an early warning system for investors, management, and regulators in anticipating bankruptcy risks. Thus, these findings confirm that SVM-based financial ratio analysis is important for supporting early detection of potential corporate bankruptcy in the Indonesian capital market.