PREDICTION OF SOFTWARE ANOMALIES METHODS BASED ON ENSEMBLE LEARNING METHODS

Software plays a vital role in all aspects of our daily lives, specifically in the fields of medicine and industry. In order to design high-quality and reliable software and avoid risks resulting from software errors, including physical and human errors, this is considered a major challenge due to t...

Full description

Saved in:
Bibliographic Details
Main Authors: Raghda Azad Hasan, Ibrahim Ahmed Saleh
Format: Article
Language:English
Published: Faculty of Engineering, University of Kufa 2025-07-01
Series:Mağallaẗ Al-kūfaẗ Al-handasiyyaẗ
Subjects:
Online Access:https://journal.uokufa.edu.iq/index.php/kje/article/view/16641
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Software plays a vital role in all aspects of our daily lives, specifically in the fields of medicine and industry. In order to design high-quality and reliable software and avoid risks resulting from software errors, including physical and human errors, this is considered a major challenge due to the limited time and budget specified. Therefore, most software development companies tend to use machine learning for prediction. With the presence of software defects that contribute to improving the quality and safety of the software produced, this is done by relying on and using records, previous projects, and available data. this paper proposed machine learning and ensemble learning suite to predict software anomalies. The evaluated approach is for models in the PROMISE real-word dataset repository containing 5 projects (Turkish company SOFTLAB). The model applies the basic algorithms (Random Forest (RF), Decision Tree (DT), Extra Tree) and the learning model ensemble (Adaboost, xgboost ,Stack, Voting, bagging) and metrics (accuracy, recall, F1 score, accuracy) to measure the prediction performance of the models and a comparison was made between the proposed model algorithms. Both adaboost , stack achieved prediction accuracy about 99.2% when implemented on the ar5 dataset.
ISSN:2071-5528
2523-0018