DEFECT SEVERITY CODE PREDICTION BASED ON ENSEMBLE LEARNING
In machine learning, learning algorithms that learn from other algorithms are called meta-learning. New algorithms called Ensemble algorithms have surfaced as a viable method to improve defect prediction models' accuracy and dependability. In software development defect prediction of software...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Lublin University of Technology
2024-12-01
|
| Series: | Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska |
| Subjects: | |
| Online Access: | https://ph.pollub.pl/index.php/iapgos/article/view/6393 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In machine learning, learning algorithms that learn from other algorithms are called meta-learning. New algorithms called Ensemble algorithms have surfaced as a viable method to improve defect prediction models' accuracy and dependability. In software development defect prediction of software engineering is still a big challenge, and leads to the failure of systems, increases the cost of maintenance, and makes the development process more difficult. Consequently, defect prediction systems have become more popular as a way to foresee possible flaws early on in the development process. Defect prediction is the process that specifies the possible defects in the code written newly or the existing modified code without the use of code testing. This paper introduces ensemble learning ideas, reviews the traditional defect prediction models, and investigates ensemble learning techniques for defect classification and prediction such as bagging, boosting, stacking, and random forests, Case studies and actual experiments illustrate the important role of ensemble algorithms in classifying five severity types of defects and predicting the severity code of defects to improve the software development process by reducing the time and effort needed to determine the type of defect.
|
|---|---|
| ISSN: | 2083-0157 2391-6761 |