A5089239701- Software Engineering Research
- Software Reliability and Analysis Research
- Software Engineering Techniques and Practices
- Software System Performance and Reliability
- Software Testing and Debugging Techniques
- Open Source Software Innovations
- Transportation and Mobility Innovations
- Bayesian Modeling and Causal Inference
- Vehicle Routing Optimization Methods
- Optimization and Search Problems
- Advanced Text Analysis Techniques
- Advanced Malware Detection Techniques
- Scientific Computing and Data Management
- Web Application Security Vulnerabilities
- Adversarial Robustness in Machine Learning
- Team Dynamics and Performance
- Computational and Text Analysis Methods
- Mobile Crowdsensing and Crowdsourcing
- Anomaly Detection Techniques and Applications
- Educational Technology and Assessment
- Engineering Education and Curriculum Development
- Data Quality and Management
- ERP Systems Implementation and Impact
- Reliability and Maintenance Optimization
- Innovation and Knowledge Management
Istanbul Technical University
2015-2023
Faculty of Media
2019
College Track
2019
University of Oulu
2012-2014
Boğaziçi University
2008-2012
Background: Most of the experiments in software engineering (SE) employ students as subjects. This raises concerns about realism results acquired through and adaptability to industry. Aim: We compare professionals understand how well represent experimental subjects SE research. Method: The comparison was made context two test-driven development conducted with an academic setting a organization. measured code quality several tasks implemented by both subject groups checked whether perform...
Abstract -- Background: Most of the experiments in software engineering (SE) employ students as subjects. This raises concerns about realism results acquired through and adaptability to industry. Aim: We compare professionals understand how well represent experimental subjects SE research. Method: The comparison was made context two test-driven development conducted with an academic setting a organization. measured code quality several tasks implemented by both subject groups checked whether...
In ICSE'08, Zimmermann and Nagappan show that network measures derived from dependency graphs are able to identify critical binaries of a complex system missed by complexity metrics. The used in their analysis is Windows product. this study, we conduct additional experiments on public data reproduce validate results. We use metrics five systems. examine three small scale embedded software two versions Eclipse compare defect prediction performance these select different granularity levels...
Software defect prediction aims to reduce software testing efforts by guiding testers through the defect‐prone sections of systems. Defect predictors are widely used in organizations predict defects order save time and effort as an alternative other techniques such manual code reviews. The usage a model real‐life setting is difficult because it requires metrics data from past projects defect‐proneness new projects. It is, on hand, very practical easy apply, can detect using less time,...
Recommendation systems in software engineering (SE) should be designed to integrate evidence into practitioners experience. Bayesian networks (BNs) provide a natural statistical framework for evidence-based decision-making by incorporating an integrated summary of the available and associated uncertainty (of consequences). In this study, we follow lead computational biology healthcare decision-making, investigate applications BNs SE terms 1) main challenges addressed, 2) techniques used...
We have conducted a study in large telecommunication company Turkey to employ software measurement program and predict pre-release defects. previously built such predictors using AI techniques. This project is transfer of our research experience into real life setting solve specific problem for the company: improve code quality by predicting defects efficiently allocating testing resources. Our results this many practical implications that managers started benefiting: analysis, bug tracking,...
Software defect data has an imbalanced and highly skewed class distribution. The misclassification costs of two classes are not equal nor known. It is critical to find the optimum bound, i.e. threshold, which would best separate defective defect-free in software data. We have applied decision threshold optimization on Naïve Bayes classifier order for ROC analyses show that significantly decreases false alarms (on average by 11%) without changing probability detection rates.
Defect prediction research mostly focus on optimizing the performance of models that are constructed for isolated projects. On other hand, recent studies try to utilize data across projects building defect models. We combine both approaches and investigate effects using mixed (i.e. within cross) project performance, which has not been addressed in previous studies. conduct experiments analyze learned from ten proprietary two different organizations. observe code metric based yield only minor...
Software vulnerabilities may lead to crucial security risks in software systems. Thus, prioritization of the is an important task for teams, and assessing how severe are would help teams during fixing maintenance activities. We replicated a prior work which aims predict severity by grouping into different levels. follow their approach on feature extraction using word embeddings, prediction model Convolutional Neural Networks (CNNs). In addition, Long Short Term Memory (LSTM) Extreme Gradient...
In this paper, we present a defect prediction model based on ensemble of classifiers, which has not been fully explored so far in type research. We have conducted several experiments public datasets. Our results reveal that classifiers considerably improve the detection capability compared to Naive Bayes algorithm. also conduct cost-benefit analysis for our ensemble, where it turns out is enough inspect 32% code average, detecting 76% defects.
Background: Writing unit tests is one of the primary activities in test-driven development. Yet, existing reviews report few evidence supporting or refuting effect this development approach on test case quality. Lack ability and skills developers to produce sufficiently good cases are also reported as limitations applying industrial practice. Objective: We investigate impact effectiveness compared an incremental last context. Method: conducted experiment setting with 24 professionals....