Software Defect Prediction for Healthcare Big Data: An Empirical Evaluation of Machine Learning Techniques

Area of Science:

• Computer Science
• Software Engineering
• Artificial Intelligence

Background:

• Software defect prediction (SDP) is crucial for ensuring software quality early in the software development life cycle (SDLC).
• Accurate defect forecasting enables development teams to optimize resource allocation and deliver high-quality software efficiently.
• Machine learning (ML) techniques offer powerful methods for identifying defective software modules by uncovering patterns in software metrics.

Purpose of the Study:

• To evaluate and compare the performance of various machine learning techniques for software defect prediction.
• To identify the most effective ML models for accurately forecasting defective software artifacts.
• To provide a benchmark for future research in the software defect prediction domain.

Main Methods:

• Utilized seven widely-used software datasets for defect prediction.
• Applied and compared ten distinct machine learning techniques: Multilayer Perceptron (MLP), Support Vector Machine (SVM), Decision Tree (J48), Radial Basis Function (RBF), Random Forest (RF), Hidden Markov Model (HMM), Credal Decision Tree (CDT), K-nearest neighbor (KNN), Average One Dependency Estimator (A1DE), and Naïve Bayes (NB).
• Evaluated model performance using metrics including Relative Absolute Error (RAE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Root Relative Squared Error (RRSE), recall, and accuracy.

Main Results:

• Random Forest (RF) achieved the highest average accuracy at 88.32%, with a rank value of 2.96.
• Support Vector Machine (SVM) demonstrated the second-best performance with 87.99% average accuracy and a rank value of 3.83.
• Credal Decision Tree (CDT) secured the third position with 87.88% average accuracy and a rank value of 3.62.

Conclusions:

• Machine learning models, particularly Random Forest and Support Vector Machine, are highly effective for software defect prediction.
• The study provides empirical evidence and benchmark results for comparing future advancements in SDP.
• These findings can guide practitioners in selecting appropriate ML techniques for enhancing software quality assurance.