Setiawan, Rudi (2025) A Hybrid Machine Learning and Optimisation-Based Model for Predicting the Success of Business-To-Consumer Software Development Projects in Indonesia. Doctoral thesis, Asia e University, Malaysia.
![[img]](http://ur.aeu.edu.my/style/images/fileicons/text.png) |
Text
Thesis Rudi Setiawan.pdf Restricted to Repository staff only Download (11MB) |
|
Text
Thesis Rudi Setiawan-1-24.pdf Download (759kB) |
Abstract
Despite the strategic importance of software projects, recent Standish Group findings indicate that only 35% achieve successful completion, with a substantial portion encountering delays or outright failure. Motivated by this challenge, this research aims to advance the predictive capability of software project success by developing an integrated artificial intelligence-based model, contextualized within B2C E-Business software projects in Indonesia. Employing a mixed-methods design, the study first identifies 28 critical success factors through empirical investigation, highlighting elements such as evaluation progress, realistic scheduling, human resource competency, project planning, and risk management. Building on these findings, a predictive framework is constructed by integrating machine learning algorithms with advanced optimization and data handling strategies. Specifically, Support Vector Machines (SVM) enhanced by Aquila Optimization demonstrate superior performance, achieving a predictive accuracy of up to 94.29% on datasets synthesized and balanced through Generative Adversarial Networks (GANs) and hybrid resampling techniques (SMOTE + Tomek Links). To translate predictive insights into actionable guidance, Shapley Additive Explanations (SHAP) are employed to quantify the contribution of each factor, enabling targeted recommendations for mitigating project risks. This study’s primary contributions lie in (i) methodologically integrating feature identification, synthetic data augmentation, imbalance handling, hyperparameter optimization, and explainable AI into a cohesive predictive pipeline; (ii) applying this integrated approach within the underexplored context of Indonesian B2C E-Business projects; and (iii) delivering practical decision-support insights that enhance project planning, thereby reducing the likelihood of costly failures. The proposed framework not only elevates predictive accuracy but also reinforces managerial strategies through data-driven prioritization of project factors, offering a robust tool to improve software project success rates in similar contexts.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Uncontrolled Keywords: | Critical success factors, software project, project management, software project prediction, software project success prediction models |
| Subjects: | H Social Sciences > HD Industries. Land use. Labor |
| Divisions: | School of Graduate Studies |
| Depositing User: | Muhamad Aizat Nazmi Mohd Nor Hamin |
| Date Deposited: | 09 Apr 2026 08:19 |
| Last Modified: | 09 Apr 2026 08:19 |
| URI: | http://ur.aeu.edu.my/id/eprint/1461 |
Actions (login required)
 |
View Item |
