Author(s)

Kufre Christopher Ukpe, Constance Izuchukwu Amannah

Department of Computer Science, Ignatius Ajuru University of Education, Rumuolumeni, Rivers State, Nigeria.

Software defect prediction and cost estimation are critical challenges in software engineering, directly influencing software quality and project management efficiency. This study presents a hybridized model integrating software defect prediction and cost estimation using machine learning techniques. Leveraging Support Vector Machines (SVM) for classification and regression, the model predicts software defects and estimates development costs with high precision. Principal Component Analysis (PCA) was employed for dimensionality reduction, ensuring computational efficiency and preserving 95% of dataset variance. The hybrid model was trained and tested on datasets sourced from the NASA PROMISE repository and Kaggle, employing a 70% - 30% train-test split with K-fold cross-validation for unbiased performance evaluation. The defect prediction component achieved an accuracy of 94.5%, precision of 93.7%, recall of 92.3%, and F1-score of 93.0%. For cost estimation, the model recorded a Coefficient of Determination (R²) of 0.87, Mean Absolute Error (MAE) of 5.2 person-hours, and Root Mean Squared Error (RMSE) of 6.8 person-hours. The proposed hybrid model outperforms traditional approaches by addressing defect detection and cost evaluation simultaneously, uncovering potential correlations between software quality attributes and cost factors. The study demonstrates the robustness of the model in real-world scenarios, providing actionable insights for improved resource allocation, reduced maintenance costs, and enhanced software reliability. Future research will explore the integration of additional machine learning techniques and extended datasets for broader applicability.

Software, Performance, Cost, Defect, Vector, Machine, Support

Ukpe, K. and Amannah, C. (2025) Hybrid Software Model for Defect Detection and Cost Evaluation Using Support Vector Machine Algorithm. Journal of Computer and Communications, 13, 244-264. doi: 10.4236/jcc.2025.134016.