Software defect prediction using semi-supervised methods

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Software defect prediction using semi-supervised methods

ارائه دهنده: Provider: Sajjad Aasal

اساتید راهنما: Supervisors: Morteza Yousef Sanati (Ph. D)

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Mehdi Sakhaei Nia (Ph. D) - Vahid Nosrati (Ph. D)

زمان و تاریخ ارائه: Time and date of presentation: 2024

مکان ارائه: Place of presentation: Faculty of Engineering

چکیده: Abstract: Accurate and timely prediction of software defects is one of the fundamental challenges in software engineering that directly affects quality, reliability, and development costs. In this research, an innovative self-supervised semi-supervised machine learning model is proposed to overcome the limitations of traditional defect prediction methods. This model has achieved significant improvements in prediction accuracy by employing intelligent data preprocessing techniques, including Gaussian distribution-based feature normalization to reduce noise and oversampling to address class imbalance. Additionally, by using a self-learning algorithm, the model can extract complex and latent patterns in software data, which is particularly important in conditions of limited labeled data. Evaluation results on the standard NASA dataset demonstrate that the proposed model, especially in terms of recall, which indicates the model's ability to identify all defective instances, outperforms traditional and unsupervised methods. This performance improvement highlights the high potential of the proposed model for early detection of software defects and reducing associated costs.

فایل: ّFile: Download فایل