Xiaoyan Zhao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiaoyan Zhao has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Xuwei Guo
Universiti Kebangsaan Malaysia, Malaysia
5 shared publications - Zulkefli Mansor
Center for Software Technology and Management, Universiti Kebangsaan Malaysia, Bangi Selangor, Malaysia
4 shared publications - Mohd Zakree Ahmad Nazri
Universiti Kebangsaan Malaysia (UKM)
3 shared publications - Liangyu Li
Universiti Kebangsaan Malaysia
2 shared publications - Liangyu Li
Universiti Kebangsaan Malaysia, Malaysia
2 shared publications - Liangyu Li
Faculty of Information Science & Technology, Universiti Kebangsaan Malaysia, Bangi Selangor, Malaysia
2 shared publications
Latest Publications
- 2 May 2026
- Published by Springer Nature in Applied Intelligence
- Vol. 56 (7) , 240
Identification and empirical investigation of factors influencing cost estimation in Agile development
- 2 May 2026
- Published by Springer Nature in Applied Intelligence
- Vol. 56 (7) , 240
Identification and empirical investigation of factors influencing cost estimation in Agile development
- Open Access
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in IET Software
- Vol. 2026 (1)
Configuration Factors for Compact LLM‐Based Story Point Estimation: A Controlled Project‐Level Empirical Study
Story point estimation is widely used in agile software planning, yet automated estimation remains challenging because issue descriptions and story point distributions vary substantially across projects. This study examines configuration sensitivity and comparative performance in compact large language model (LLM)‐based story point estimation through a controlled project‐level study using five TAWOS projects and two ~1–1.5B‐parameter compact backbones with frozen parameters and trainable prediction heads. The study comprises four‐factor configuration analysis covering maximum sequence length, metadata concatenation, learning rate, and soft‐target mixing; metadata‐free sensitivity analysis; learning‐rate–soft‐target‐mixing interaction analysis; comparison with three aligned baselines; assessment of alternative temperature‐selection objectives; and computational‐cost evaluation. Soft‐target mixing and learning rate show the larger average mean absolute error (MAE) effects among the four evaluated factors, with mean high‐minus‐low effects of −0.0535 and −0.0458, respectively, although both vary in direction across project–backbone–seed scenarios. The metadata‐free analysis shows similar average effect magnitudes for learning rate and soft‐target mixing, while the interaction between these two factors is heterogeneous, with 11 negative and 9 positive scenario‐level contrasts. When compact configurations and baseline methods are selected independently using seed‐averaged validation MAE, the corresponding test results favor the compact configuration in three projects and the selected baseline in two, with additional project‐specific variation among TFIDF‐LGBM, MiniLM‐MLP, and CodeBERT‐FT. Temperature‐selection rules produce different MAE, snapped‐expected calibration error (ECE), and negative log‐likelihood (NLL) profiles. The compact configurations require very small trainable prediction heads but retain substantially higher inference latency and memory requirements than the evaluated baselines; method‐level mean raw‐text‐to‐logits latency ranges ~0.37–25.72 ms/sample in the reported NVIDIA A10 environment. Overall, the results characterize compact‐LLM story point estimation by heterogeneous configuration sensitivity, project‐dependent model ranking, and accuracy–cost trade‐offs rather than a single consistently preferable configuration or model family.
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in IET Software
- Vol. 2026 (1)
Open AccessConfiguration Factors for Compact LLM‐Based Story Point Estimation: A Controlled Project‐Level Empirical Study
Story point estimation is widely used in agile software planning, yet automated estimation remains challenging because issue descriptions and story point distributions vary substantially across projects. This study examines configuration sensitivity and comparative performance in compact large language model (LLM)‐based story point estimation through a controlled project‐level study using five TAWOS projects and two ~1–1.5B‐parameter compact backbones with frozen parameters and trainable prediction heads. The study comprises four‐factor configuration analysis covering maximum sequence length, metadata concatenation, learning rate, and soft‐target mixing; metadata‐free sensitivity analysis; learning‐rate–soft‐target‐mixing interaction analysis; comparison with three aligned baselines; assessment of alternative temperature‐selection objectives; and computational‐cost evaluation. Soft‐target mixing and learning rate show the larger average mean absolute error (MAE) effects among the four evaluated factors, with mean high‐minus‐low effects of −0.0535 and −0.0458, respectively, although both vary in direction across project–backbone–seed scenarios. The metadata‐free analysis shows similar average effect magnitudes for learning rate and soft‐target mixing, while the interaction between these two factors is heterogeneous, with 11 negative and 9 positive scenario‐level contrasts. When compact configurations and baseline methods are selected independently using seed‐averaged validation MAE, the corresponding test results favor the compact configuration in three projects and the selected baseline in two, with additional project‐specific variation among TFIDF‐LGBM, MiniLM‐MLP, and CodeBERT‐FT. Temperature‐selection rules produce different MAE, snapped‐expected calibration error (ECE), and negative log‐likelihood (NLL) profiles. The compact configurations require very small trainable prediction heads but retain substantially higher inference latency and memory requirements than the evaluated baselines; method‐level mean raw‐text‐to‐logits latency ranges ~0.37–25.72 ms/sample in the reported NVIDIA A10 environment. Overall, the results characterize compact‐LLM story point estimation by heterogeneous configuration sensitivity, project‐dependent model ranking, and accuracy–cost trade‐offs rather than a single consistently preferable configuration or model family.
- Open Access
- 1 January 2026
- Published by IGI Global Scientific Publishing in International Journal of Information Technologies and Systems Approach
- Vol. 19 (1) , 1-23
An Approach to Cost Estimation in Medical Software Based on Improved Genetic Algorithm
Accurate estimation of medical software development costs is critical for project planning and resource management. Conventional genetic algorithms often show limited efficiency and suboptimal accuracy on large-scale, heterogeneous datasets. To address these issues, the authors propose an enhanced genetic algorithm framework that integrates the AO optimizer with the Chernobyl disaster optimizer (CDO). The AO optimizer supports global search and feature extraction of cost parameters, while the CDO enhances prediction precision and helps avoid local optima. Key cost-related variables were identified and assessed using logistic regression, with results visualized via survival and forest plots. This dual-optimizer design improves convergence speed, adaptability, and predictive reliability. Experiments confirm that the model achieves higher sensitivity and accuracy compared to traditional methods. Overall, the approach provides a robust and scalable tool for medical software cost estimation, highlighting the value of hybrid intelligent optimization in medical informatics.
- 1 January 2026
- Published by IGI Global Scientific Publishing in International Journal of Information Technologies and Systems Approach
- Vol. 19 (1) , 1-23
Open AccessAn Approach to Cost Estimation in Medical Software Based on Improved Genetic Algorithm
Accurate estimation of medical software development costs is critical for project planning and resource management. Conventional genetic algorithms often show limited efficiency and suboptimal accuracy on large-scale, heterogeneous datasets. To address these issues, the authors propose an enhanced genetic algorithm framework that integrates the AO optimizer with the Chernobyl disaster optimizer (CDO). The AO optimizer supports global search and feature extraction of cost parameters, while the CDO enhances prediction precision and helps avoid local optima. Key cost-related variables were identified and assessed using logistic regression, with results visualized via survival and forest plots. This dual-optimizer design improves convergence speed, adaptability, and predictive reliability. Experiments confirm that the model achieves higher sensitivity and accuracy compared to traditional methods. Overall, the approach provides a robust and scalable tool for medical software cost estimation, highlighting the value of hybrid intelligent optimization in medical informatics.
- Open Access
- 23 June 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (5) , 1-21
A data-driven cost estimation model for agile development based on Kolmogorov-Arnold Networks and AdamW optimization
- 23 June 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (5) , 1-21
Open AccessA data-driven cost estimation model for agile development based on Kolmogorov-Arnold Networks and AdamW optimization
- 29 May 2025
- Published by SAGE Publications in Intelligent Data Analysis: An International Journal
- Vol. 30 (1) , 217-234
A review of optimization techniques for cost estimation in software development
In software development, cost estimation remains a significant challenge. Despite numerous research efforts, identifying an optimal technique applicable to all situations has proven difficult. The increasing adoption of agile development methods has further complicated accurate cost estimation. Recently, the application of optimization techniques, particularly metaheuristic optimization algorithms, has increased to enhance estimation accuracy and performance. However, there is a lack of systematic literature reviews exploring these optimization techniques in software cost estimation (SCE). This study aims to fill this gap by employing a systematic literature review (SLR) method to select, filter, and analyze relevant literature from 2019 to 2024. The review included 41 journal articles and 11 conference proceedings to evaluate the current development status of optimization methodologies in SCE. The study identified 20 key optimization algorithms with the top 6 most commonly used: Particle Swarm Optimization (PSO), Differential Evolution (DE), Genetic Algorithm (GA), Gray Wolf Optimization (GWO), Whale Optimization Algorithm (WOA), and Bat algorithm (BA). Their key contributions are parameter tuning, feature analysis, and model optimization. While optimization techniques hold promise, their application in cost estimation must carefully consider these challenges, such as premature convergence, sensitivity to initial parameters, etc. Findings also reveal significant limitations for Agile Software Cost Estimation (ASCE), such as the less applicability of traditional cost estimation techniques, lack of cost drivers for agile, more reliance on expertise and experience, and few public datasets. This research provides valuable insights for further exploration of optimization techniques in SCE.
- 29 May 2025
- Published by SAGE Publications in Intelligent Data Analysis: An International Journal
- Vol. 30 (1) , 217-234
A review of optimization techniques for cost estimation in software development
In software development, cost estimation remains a significant challenge. Despite numerous research efforts, identifying an optimal technique applicable to all situations has proven difficult. The increasing adoption of agile development methods has further complicated accurate cost estimation. Recently, the application of optimization techniques, particularly metaheuristic optimization algorithms, has increased to enhance estimation accuracy and performance. However, there is a lack of systematic literature reviews exploring these optimization techniques in software cost estimation (SCE). This study aims to fill this gap by employing a systematic literature review (SLR) method to select, filter, and analyze relevant literature from 2019 to 2024. The review included 41 journal articles and 11 conference proceedings to evaluate the current development status of optimization methodologies in SCE. The study identified 20 key optimization algorithms with the top 6 most commonly used: Particle Swarm Optimization (PSO), Differential Evolution (DE), Genetic Algorithm (GA), Gray Wolf Optimization (GWO), Whale Optimization Algorithm (WOA), and Bat algorithm (BA). Their key contributions are parameter tuning, feature analysis, and model optimization. While optimization techniques hold promise, their application in cost estimation must carefully consider these challenges, such as premature convergence, sensitivity to initial parameters, etc. Findings also reveal significant limitations for Agile Software Cost Estimation (ASCE), such as the less applicability of traditional cost estimation techniques, lack of cost drivers for agile, more reliance on expertise and experience, and few public datasets. This research provides valuable insights for further exploration of optimization techniques in SCE.
- Open Access
- 7 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 63219-63236
Advancing Agile Software Cost Estimation Through Data Synthesis: A Comparative Analysis of Five Generation Techniques
Agile has been used in software development for over 20 years and is the preferred development method for more than 85% of software companies. However, cost estimation in agile development remains a significant challenge. This is reflected in the fact that the accuracy of estimation still needs to be improved, and most cost estimation techniques still depend on the experience and knowledge of the team. While machine learning algorithms have performed better in this area, the lack of sufficient agile cost data hinders large-scale training and in-depth research. To address this issue, this study selected five data generation techniques—Variational Autoencoder (VAE), Wasserstein Generative Adversarial Network (WGAN), Synthetic Minority Over-sampling Technique for Nominal and Continuous Features (SMOTE-NC), Data Augmentation for Tabular Data (Augmentation), and Tabular Data Diffusion Probabilistic Models (TabDDPM)—based on the characteristics of agile cost data. Using cost data from 75 agile projects, these techniques were employed to generate three sets of data with sizes of 200, 500, and 1000. A performance evaluation model was created based on consistency, authenticity, diversity, and effectiveness to verify the performance of these generated data. The experimental results show that WGAN consistently scored 16 out of 20 points across all three data sets, excelling in data consistency and authenticity. SMOTE-NC and Augmentation followed. SMOTE-NC scored 15 out of 20 points for all data sizes and performed best in effectiveness, with an MMRE of 88.16% and a PRED (0.2) of 84.5%. Augmentation had the highest performance when generating 1000 data points. These findings highlight the potential of data generation technologies, especially WGAN, in improving agile cost estimation and providing guidance on selecting appropriate data amounts. This lays a foundation for further development of machine learning algorithms in this field and offers valuable insights for other researchers.
- 7 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 63219-63236
Open AccessAdvancing Agile Software Cost Estimation Through Data Synthesis: A Comparative Analysis of Five Generation Techniques
Agile has been used in software development for over 20 years and is the preferred development method for more than 85% of software companies. However, cost estimation in agile development remains a significant challenge. This is reflected in the fact that the accuracy of estimation still needs to be improved, and most cost estimation techniques still depend on the experience and knowledge of the team. While machine learning algorithms have performed better in this area, the lack of sufficient agile cost data hinders large-scale training and in-depth research. To address this issue, this study selected five data generation techniques—Variational Autoencoder (VAE), Wasserstein Generative Adversarial Network (WGAN), Synthetic Minority Over-sampling Technique for Nominal and Continuous Features (SMOTE-NC), Data Augmentation for Tabular Data (Augmentation), and Tabular Data Diffusion Probabilistic Models (TabDDPM)—based on the characteristics of agile cost data. Using cost data from 75 agile projects, these techniques were employed to generate three sets of data with sizes of 200, 500, and 1000. A performance evaluation model was created based on consistency, authenticity, diversity, and effectiveness to verify the performance of these generated data. The experimental results show that WGAN consistently scored 16 out of 20 points across all three data sets, excelling in data consistency and authenticity. SMOTE-NC and Augmentation followed. SMOTE-NC scored 15 out of 20 points for all data sizes and performed best in effectiveness, with an MMRE of 88.16% and a PRED (0.2) of 84.5%. Augmentation had the highest performance when generating 1000 data points. These findings highlight the potential of data generation technologies, especially WGAN, in improving agile cost estimation and providing guidance on selecting appropriate data amounts. This lays a foundation for further development of machine learning algorithms in this field and offers valuable insights for other researchers.
- Open Access
- 1 January 2024
- Published by IGI Global Scientific Publishing in International Journal of Information Technologies and Systems Approach
- Vol. 17 (1) , 1-22
Medical Software Engineering Cost Estimation Model Based on Multimodal Improved Genetic Algorithm
Cost estimation is crucial in the development of medical software and is typically achieved through the integration of cost data and the application of AI algorithms. Traditional methods, such as unoptimized machine learning, struggle with inefficiencies and low accuracy in processing large volumes of dynamic data. To address these issues, the present study has developed an enhanced cost estimation model for medical software engineering using an mutimodal improved genetic algorithm. This model aims to enhance responsiveness, adaptability, and accuracy. We employed a COX proportional hazards regression model to analyze factors influencing development costs. Additionally, we integrated an Arithmetic Optimization Algorithm (AOA) with a Dwarf Mongoose Optimizer (DMOA) to identify cost features using AOA-GA and to predict costs using DMOA-GA. The results indicate enhanced sensitivity to data input changes, high accuracy, and stability, effectively facilitating cost estimation in complex environments for medical software projects.
- 1 January 2024
- Published by IGI Global Scientific Publishing in International Journal of Information Technologies and Systems Approach
- Vol. 17 (1) , 1-22
Open AccessMedical Software Engineering Cost Estimation Model Based on Multimodal Improved Genetic Algorithm
Cost estimation is crucial in the development of medical software and is typically achieved through the integration of cost data and the application of AI algorithms. Traditional methods, such as unoptimized machine learning, struggle with inefficiencies and low accuracy in processing large volumes of dynamic data. To address these issues, the present study has developed an enhanced cost estimation model for medical software engineering using an mutimodal improved genetic algorithm. This model aims to enhance responsiveness, adaptability, and accuracy. We employed a COX proportional hazards regression model to analyze factors influencing development costs. Additionally, we integrated an Arithmetic Optimization Algorithm (AOA) with a Dwarf Mongoose Optimizer (DMOA) to identify cost features using AOA-GA and to predict costs using DMOA-GA. The results indicate enhanced sensitivity to data input changes, high accuracy, and stability, effectively facilitating cost estimation in complex environments for medical software projects.