Scholarlink

<p>The present study investigates machine-learning approaches to predicting financial distress among Indonesian firms using annual data for 325 companies (2013–2023 3575 observations). The feature set comprises financial and operational predictors, in addition to a binary outcome indicator of distress. In order to mitigate severe class imbalance driven by the rarity of distress events, the Synthetic Minority Over-sampling Technique (SMOTE) is applied within training folds, while the validation and test sets are left untouched in order to avoid leakage. Principal Component Analysis (PCA) facilitates exploratory visualisation. The performance of Support Vector Machine (SVM), Decision Tree, Random Forest, k-Nearest Neighbors (KNN), Gradient Boosting, and a deep neural network (ANN/MLP) is benchmarked under a stratified evaluation protocol. Among conventional models, the Decision Tree demonstrates the strongest F1 performance (0.8980), followed closely by Gradient Boosting (0.8627) and Random Forest (0.8462). In the case of the deep model, the best configuration (50 neurons, Sigmoid activation, 100 epochs, batch size 8) achieved 99.20% accuracy and an F1-score of 0.95, indicating robust minority-class detection alongside high overall correctness. The findings underscore the effectiveness of ensemble learners and optimally tuned deep neural networks in predicting financial distress in the context of emerging markets. The study’s practical relevance aligns with the Development Goal (SDG) 8: Decent Work and Economic Growth, the earlier and more reliable detection of financial distress can help firms take preventive actions that reduce job losses and operational disruptions, Allowing businesses to predict and act upon their financial challenges, this study nurtures resilience and sustainable development within the corporate sector, the main driver of economic activities. Besides, addressing financial distress in an emerging market like Emerging Markets contributes to inclusive economic development through stability that helps workers, shareholders, and communities dependent on such businesses.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Financial distress prediction in emerging markets by using a machine learning approach for financial sustainability

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月24日

查看全文
  • Farida Titik Kristanti,
  • Dwi Fitrizal Salim,
  • Aditya Firman Ihsan

摘要

The present study investigates machine-learning approaches to predicting financial distress among Indonesian firms using annual data for 325 companies (2013–2023 3575 observations). The feature set comprises financial and operational predictors, in addition to a binary outcome indicator of distress. In order to mitigate severe class imbalance driven by the rarity of distress events, the Synthetic Minority Over-sampling Technique (SMOTE) is applied within training folds, while the validation and test sets are left untouched in order to avoid leakage. Principal Component Analysis (PCA) facilitates exploratory visualisation. The performance of Support Vector Machine (SVM), Decision Tree, Random Forest, k-Nearest Neighbors (KNN), Gradient Boosting, and a deep neural network (ANN/MLP) is benchmarked under a stratified evaluation protocol. Among conventional models, the Decision Tree demonstrates the strongest F1 performance (0.8980), followed closely by Gradient Boosting (0.8627) and Random Forest (0.8462). In the case of the deep model, the best configuration (50 neurons, Sigmoid activation, 100 epochs, batch size 8) achieved 99.20% accuracy and an F1-score of 0.95, indicating robust minority-class detection alongside high overall correctness. The findings underscore the effectiveness of ensemble learners and optimally tuned deep neural networks in predicting financial distress in the context of emerging markets. The study’s practical relevance aligns with the Development Goal (SDG) 8: Decent Work and Economic Growth, the earlier and more reliable detection of financial distress can help firms take preventive actions that reduce job losses and operational disruptions, Allowing businesses to predict and act upon their financial challenges, this study nurtures resilience and sustainable development within the corporate sector, the main driver of economic activities. Besides, addressing financial distress in an emerging market like Emerging Markets contributes to inclusive economic development through stability that helps workers, shareholders, and communities dependent on such businesses.