Academic Journal
Optimized Software Defect Prediction using SMOTE-Augmented Random Forest on High-Dimensional Code Metrics.
| Title: | Optimized Software Defect Prediction using SMOTE-Augmented Random Forest on High-Dimensional Code Metrics. |
|---|---|
| Authors: | Yudhistiro, Kukuh |
| Source: | Internet of Things & Artificial Intelligence Journal (IOTA); May2026, Vol. 6 Issue 2, p172-184, 13p |
| Subject Terms: | Random forest algorithms, Resampling (Statistics), Defect tracking (Computer software development), Machine learning, Software measurement |
| Abstract: | Software Defect Prediction (SDP) is crucial for improving software quality and reducing maintenance costs. Class imbalance in SDP datasets often leads to biased models favoring the majority (non-defective) class. This paper proposes a hybrid approach combining the Synthetic Minority Over-sampling Technique (SMOTE) with Random Forest (RF) to address imbalance and enhance prediction accuracy. Using a custom dataset of 500 software modules characterized by 35 source code metrics (34 features + defect label, e.g., Average Blank Lines - abl, McCabe's Cyclomatic Complexity - mcc), we compare our method against baselines: Naive Bayes (NB) and Logistic Regression (LR). Results show our SMOTE+RF achieves perfect performance (Accuracy=0.93, Recall=0.91, F1-Score=0.92, AUC=0.97), outperforming NB (Accuracy=0.95, Recall=0.92) and LR (Accuracy=0.95, Recall=0.95). Compared to baseline studies using vanilla ML algorithms on NASA datasets, where RF accuracies range from 80-90% without imbalance handling, our approach demonstrates superior handling of imbalance and higher defect detection rates. Feature importance analysis highlights metrics like Halstead Total Operands (n) and McCabe's Cyclomatic Complexity (mcc) as key predictors. [ABSTRACT FROM AUTHOR] |
| Copyright of Internet of Things & Artificial Intelligence Journal (IOTA) is the property of Internet of Things & Artificial Intelligence Journal (IOTA) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Complementary Index |
Be the first to leave a comment!