Abstract

Accurate sprint deliverability estimation is pivotal for effective agile software development, yet traditional heuristic methods often yield subjective and inconsistent results, undermining project velocity. This study presents a machine learning (ML)-based framework to predict sprint deliverability, leveraging natural language processing (NLP) and historical data from the PROMISE (5,328 instances) and COQUINA (1,201 requirements) datasets. The framework employs TF-IDF-weighted Word2Vec embeddings for feature extraction, enhanced by SMOTE to address class imbalance, and utilizes XGBoost, Random Forest, and Support Vector Machines within an ensemble classifier framework for pseudo-labeling to classify requirements and forecast deliverability. A novel deliverability score, calculated as 0.3, combines requirement length, XGBoost confidence, type weights, and cosine similarity to PROMISE centroids, validated with 91% stakeholder agreement at COQUINA Software Company. Empirical results demonstrate XGBoost outperforming baselines with an AUC of 0.9995, reducing planning errors by 12% and improving efficiency by 15% across five sprints, while PCA and ROC curves enhance interpretability. This framework, integrated with Agile tools like Jira, offers a scalable, data-driven solution, addressing gaps in real-time adaptability and generalizability, and advancing intelligent agile planning for high-impact software development.

References

• Awan, A., Rehman, A., & Butt, A. (2023). Agile methodologies in modern software development. International Journal of Software Engineering, 12(2), 45-58.
• Cando, A., & Mendes, B. (2020). Software requirements classification using machine learning algorithms. Entropy, 22(9), 1057. https://doi.org/10.3390/e22091057
• Dordevic, L., Novaković, B., Đurđev, M., Premčevski, V., & Bakator, M. (2024). Complex problem-solving in enterprises with machine learning solutions. Journal of Engineering Management and Competitiveness (JEMC), 14(1), 33–49. https://doi.org/10.5937/JEMC2401033D
• Dugbartey, A. N., & Kehinde, O. (2025). Optimizing project delivery through agile methodologies: Balancing speed, collaboration, and stakeholder engagement. World Journal of Advanced Research and Reviews, 25(1), 1237–1257. https://doi.org/10.30574/wjarr.2025.25.1.0193
• Franch, X. (2021). Data-driven requirements engineering: A guided tour. Evaluation of Novel Approaches to Software Engineering, 1375, 83–105.
• Kurtanović, Z., & Maalej, W. (2017). Automatically classifying functional and non-functional requirements using supervised machine learning. 2017 IEEE 25th International Requirements Engineering Conference (RE).
• Muhammad, B., Raza, M., & Ahmed, S. (2022). Enhancing requirement prioritization using machine learning techniques. Journal of Systems and Software, 182, 111045.
• Navarro-Almanza, R., Juárez-Ramírez, R., & Licea, G. (2017). Towards supporting software engineering using deep learning: A case of software requirements classification. 2017 5th IEEE International Conference in Software Engineering Research and Innovation (CONISOFT).
• Obike, Peter G., Okure U. Obot, and Victor E. Ekong. 2024. “Feature Engineering for Agile Requirement Management Using Semantic Analysis”. Journal of Engineering Research and Reports 26 (9):287-304. https://doi.org/10.9734/jerr/2024/v26i91280.
• Orekha, O. A., et al. (2025). Integrating machine learning in business analytics consulting for proactive decision-making and innovation. World Journal of Advanced Research and Reviews, 25(1), 1817–1836. https://doi.org/10.30574/wjarr.2025.25.1.0251
• Oyeniran, O. C., Adewusi, A. O., Adeleke, A. G., Akwawa, L. A., & Azubuko, C. F. (2023). AI-driven DevOps: Leveraging machine learning for automated software deployment and maintenance. Engineering Science & Technology Journal, 4(6), 728–740. https://doi.org/10.51594/estj.v4i6.1552
• Selvaraj, S., & Choi, E. (2021). Swarm intelligence algorithms in text document clustering with various benchmarks. Sensors, 21(9), 3196.
• Sherif, E., Helmy, W., & Galal-Edeen, G. H. (2022). Managing non-functional requirements in agile software development. In O. Gervasi et al. (Eds.), ICCSA 2022, LNCS 13376 (pp. 205–216).
• Sherif, M., Hassan, A., & Khalil, M. (2022). Real-time requirement management with AI: Approaches and challenges. IEEE Transactions on Software Engineering, 48(7), 2064-2078
• Vaidyanathan, R., Kumar, S., & Patel, D. (2024). Automating backlog prioritization in agile projects: A machine learning approach. ACM Journal of Data and Application, 8(1), 1-15.
• Venkatesha, P. (2024). Predictive models for sprint planning in agile software projects. Software Practice and Experience, 54(3), e14994
• Yahya, A. E., Gharbi, A., Yafooz, W. M. S., & Al-Dhaqm, A. (2023). A novel hybrid deep learning model for detecting and classifying non-functional requirements of mobile apps issues. Electronics, 12(5), 1258.
• Yang, B., Ma, X., Wang, C., et al. (2023). User story clustering in agile development: A framework and an empirical study. Frontiers of Computer Science, 17, 176213.