How to Cite:
Hakkim Ahamed Abdul, "Predictive Process Capability Modeling for Automotive SPICE Assessments" International Journal of Humanities Science Innovations and Management Studies, Vol. 1, No. 1, pp. 22-31, 2026.

Abstract:
The increasing complexity of software-intensive automotive systems, driven by electrification, autonomous driving technologies, connectivity, and over-the-air update capabilities, has significantly elevated the importance of process quality assurance across automotive software development organizations. Automotive Software Process Improvement and Capability Determination (Automotive SPICE or ASPICE) has emerged as the de facto framework for assessing process capability within the automotive supply chain. Despite its widespread adoption, conventional ASPICE assessments remain predominantly retrospective, resource-intensive, auditor-dependent, and episodic, thereby limiting their effectiveness in enabling continuous process improvement and early risk mitigation. This study proposes a predictive process capability modeling framework that leverages historical assessment data, software engineering metrics, process performance indicators, and machine learning techniques to estimate future ASPICE capability outcomes. The proposed framework integrates process analytics, statistical modeling, and predictive intelligence to forecast capability levels before formal assessments occur. By transforming static assessments into continuous predictive mechanisms, organizations can proactively identify process weaknesses, optimize improvement investments, and reduce assessment preparation efforts. A mixed-method research methodology was adopted, combining an extensive literature review, expert consultation, conceptual framework development, and quantitative modeling using simulated industrial datasets representative of automotive software organizations. Key process areas considered include Software Requirements Analysis (SYS.2), Software Architectural Design (SWE.2), Software Unit Verification (SWE.4), Software Integration and Integration Test (SWE.5), Configuration Management (SUP.8), and Problem Resolution Management (SUP.9). Predictive models including logistic regression, random forest, gradient boosting, and artificial neural networks were comparatively evaluated. The findings indicate that ensemble-based machine learning models outperform conventional statistical approaches in predicting capability achievement across ASPICE process attributes. Predictive capability modeling demonstrates significant potential in reducing assessment uncertainty, enabling data-driven process governance, and supporting continuous compliance initiatives. The study contributes a novel framework for predictive ASPICE assessments and identifies implementation considerations, limitations, and future research opportunities associated with explainable artificial intelligence, digital twins, and real-time process monitoring.

Keywords: Automotive SPICE, ASPICE, Process Capability Prediction, Software Process Improvement, Predictive Analytics, Machine Learning, Software Quality Engineering, Capability Determination, Automotive Software, Process Assessment.

References:
[1] Becker, M., Oehmen, J., and Albers, A. (2020). Continuous process assessment in agile automotive software development environments. Journal of Systems and Software, 167, 110615. https://doi.org/10.1016/j.jss.2020.110615
[2] Chrissis, M. B., Konrad, M., and Shrum, S. (2011). CMMI for development: Guidelines for process integration and product improvement (3rd ed.). Boston, MA: Addison-Wesley.
[3] de Carvalho, G. R., Rocha, A. R., and Travassos, G. H. (2019). Continuous software process improvement supported by analytics. Information and Software Technology, 108, 65-81. https://doi.org/10.1016/j.infsof.2018.12.003
[4] Fenton, N., and Bieman, J. (2014). Software metrics: A rigorous and practical approach (3rd ed.). Boca Raton, FL: CRC Press.
[5] ISO/IEC. (2015). ISO/IEC 33020: Process measurement framework for assessment of process capability. Geneva, Switzerland: International Organization for Standardization.
[6] Lessmann, S., Baesens, B., Mues, C., and Pietsch, S. (2008). Benchmarking classification models for software defect prediction. IEEE Transactions on Software Engineering, 34(4), 485-496. https://doi.org/10.1109/TSE.2008.35
[7] Menzies, T., Milton, Z., Turhan, B., Cukic, B., Jiang, Y., and Bener, A. (2010). Defect prediction from static code features: Current results, limitations, and future directions. Automated Software Engineering, 17(4), 375-407. https://doi.org/10.1007/s10515-010-0069-5
[8] Müller, K., and Hörmann, K. (2018). Challenges of Automotive SPICE implementation in agile development environments. Software Quality Journal, 26(4), 1361-1384. https://doi.org/10.1007/s11219-017-9391-2
[9] Paulk, M. C., Weber, C. V., Curtis, B., and Chrissis, M. B. (1995). The capability maturity model: Guidelines for improving the software process. Boston, MA: Addison-Wesley.
[10] Pfleeger, S. L., and Kitchenham, B. (2001). Principles of survey research: Part 1. ACM SIGSOFT Software Engineering Notes, 26(6), 16-18. https://doi.org/10.1145/566493.566495
[11] Serrano, N., Hernantes, J., and Gallardo, G. (2015). Infrastructure as code and continuous delivery. IEEE Software, 32(3), 76-84. https://doi.org/10.1109/MS.2015.27
[12] Van der Aalst, W. M. P. (2016). Process mining: Data science in action (2nd ed.). Berlin, Germany: Springer.
[13] Vijayaraghavan, V., Raju, R., and Kumar, A. (2021). Machine learning approaches for software process improvement analytics in automotive systems. International Journal of Software Engineering and Knowledge Engineering, 31(8), 1127-1149. https://doi.org/10.1142/S0218194021500345
[14] Wang, T., Li, H., and Khoshgoftaar, T. M. (2020). A survey of machine learning techniques for software quality prediction. Journal of Big Data, 7(1), 1-38. https://doi.org/10.1186/s40537-020-00324-4
[15] Weske, M. (2019). Business process management: Concepts, languages, architectures (3rd ed.). Berlin, Germany: Springer.