NATURE BASED PREDICTION MODEL OF BUG REPORTS BASED ON ENSEMBLE MACHINE LEARNING MODEL

Abstract

Software bug reports are essential for identifying, tracking, prioritizing, and resolving faults during software development and maintenance. Accurate prediction of bug reports can help developers detect defects, reduce debugging effort, and improve software quality. This work presents a NatureBased Prediction Model of Bug Reports based on Ensemble Machine Learning. The proposed framework integrates data preprocessing, feature extraction, nature-based feature selection, and ensemble classification. Initially, bug report data are cleaned by handling missing values, duplicate records, noise, and inconsistent information. A nature-based optimization technique selects informative features while eliminating redundant attributes and reducing computational complexity. The optimized features are provided to Random Forest, Gradient Boosting, and XGBoost classifiers, whose predictions are combined using ensemble learning. The model is evaluated using accuracy, precision, recall, F1-score, and ROC-AUC. The proposed approach aims to improve prediction performance, reduce false predictions, support early fault detection, minimize maintenance costs, and enhance software stability, scalability, and reliability. Keywords: Bug Prediction, Ensemble Learning, Nature-Based Optimization, Random Forest, XGBoost, Software Defects.