Technical Debt Analysis In Software Systems: Implications For System Performance And Maintainability

Abstract

Technical debt (TD) is a persistent challenge in software engineering that impacts system performance, maintainability, and long-term development cost. It arises from suboptimal design decisions, rushed implementations, and evolving requirements that gradually degrade software quality. Existing studies lack a comprehensive model for analyzing TD at the component level and linking it to maintainability degradation and performance loss. This research proposes the Tasmanian Devil mutated Weighted Random Forest (TD-WRF) model to improve understanding and prediction of maintainability outcomes influenced by accumulated TD. The WRF model assigns adaptive weights to features and decision trees to prioritize key TD indicators, while Tasmanian Devil optimization applies controlled stochastic perturbation to enhance pattern exploration and reduce overfitting. A comprehensive dataset consisting of technical debt indicators such as code smells, complexity metrics, and architectural violations is utilized for experimentation. Min-Max normalization is applied during data preprocessing to scale all features uniformly and reduce bias caused by differing metric ranges. Independent Component Analysis (ICA) is used for feature extraction to obtain statistically independent and informative components for model training.The results show that accumulated technical debt increases refactoring effort, reduces readability, and raises maintenance costs, while comparative evaluation demonstrates that the proposed model achieves strong performance with a precision of 0.934. The system is implemented using Python with Scikit-learn and NumPy libraries, ensuring efficient model development and evaluation. In conclusion, the proposed TD analysis model provides a reliable and scalable approach for understanding and predicting the impact of technical debt on software maintainability and system quality.