Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{197426,
        author = {Ms. Neha Shireen and Mr. Mohammed Fahad Nawaz Khan and Mr. Mohammed Aleemuddin Ahmed and Mr. Fazal Ur Rahman and Mr. Abdul Fattah},
        title = {Nature-Based Bug Report Prediction Using Ensemble Machine Learning with Text Augmentation and XGBoost Extension},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {6293-6303},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=197426},
        abstract = {Software maintenance has evolved into one of the most resource-intensive phases of the software development lifecycle, primarily because of the continuously increasing volume of bug reports generated from large-scale software products. Modern bug repositories such as Mozilla and Eclipse accumulate thousands of reports containing textual descriptions, metadata, severity indicators, and reproduction steps. Manual analysis of these reports for identifying the nature of the bug is time-consuming, error-prone, and operationally expensive. Existing research has focused mainly on severity prediction, priority estimation, duplicate detection, and bug localization, while comparatively limited attention has been given to nature-based multiclass classification.
This research proposes a text-heavy ensemble machine learning framework for nature-based bug report prediction that combines Natural Language Processing (NLP), TF-IDF feature extraction, text augmentation, multiple baseline classifiers, and an advanced voting ensemble mechanism. The proposed system uses Support Vector Machine, Random Forest, and Logistic Regression as foundational classifiers and combines them through hard and soft voting
 strategies. To further strengthen predictive capability, an XGBoost extension layer is introduced as the advanced experimental contribution of this work. The model is evaluated on benchmark Eclipse and Mozilla bug repositories containing six bug nature classes: Program Anomaly, GUI, Network or Security, Configuration, Performance, and Test-Code.
Base classifiers such as Support Vector Machine, Random Forest, and Logistic Regression are combined using a voting ensemble classifier to improve prediction robustness. As an extension, XGBoost is incorporated to further enhance classification accuracy. Experimental results show that standalone classifiers achieve 70-77% accuracy, the voting ensemble achieves 89%, and the XGBoost extension achieves the best performance with 92% accuracy.},
        keywords = {Bug Report Classification, Ensemble Learning, Natural Language Processing, TF-IDF, XGBoost, Software Maintenance, Text Augmentation, Multiclass Classification.},
        month = {April},
        }