ITS-SGBM: A Hybrid Intelligent Tuna Swarm Optimization and Stochastic Gradient Boosting Machine Model for Financial Performance Forecasting

Abstract

In the evolving financial landscape, precise performance measurement is crucial for informed decision-making in accounting systems. Conventional approaches often struggle to handle complex business processes and regulatory changes. The research suggests a hybrid Intelligent Tuna Swarm Optimized Stochastic Gradient Boosting Machine (ITS-SGBM) model for improving financial performance assessment. Financial data from balance sheets and annual reports (2009–2022) were pre-processed through normalization, and Principal Component Analysis (PCA) was extracted as features. The dataset was split into training and testing sets, and ITS-SGBM parameters were enhanced using Intelligent Tuna Swarm Optimization to balance exploration and exploitation during model training. Model performance was confirmed using k-fold cross-validation and benchmarked against remaining models, including DGRU-IMPA and DNN-OSIE-CHOA. In addition, the TOPSIS decision-making framework was functional to rank firms based on expert-defined financial criteria. Outcomes show that the offered model accomplished superior forecasting accuracy (96.2%) with lower error rates (MAE: 0.009, RMSE: 0.005, RRMSE: 0.072) and ranking consistency of 0.94, outperforming the benchmark models. The findings confirm that the ITS-SGBM–TOPSIS incorporation provides a strong, data-driven decision-support mechanism, increasing financial stability, strategic planning, and sustainable business development.